Is 6 inches the standard for all speakers may they be full range , two way or three way?
Once a suitable 6 inch driver is chosen , a suitable one inch tweeter is added to the design. Munson showed that were more concerned with the 125Hz to 16kHz listening range which is effectively covered by a two way design through careful selection of a 6 inch driver and a 1 inch driver. Full rangers argue there's no content at 16kHz and choose a single 6 inch driver placed within a suitable enclosure that minimizes cone movement to control modulation. This summarizes speaker design.
Anything after 125Hz requires power, surface area, excursion. To be modest in two way just add another 6 inch driver, popular with Klipsch. With more amplifier power one can go three way by adding two 8 inch or two 10 inch drivers with bass reflex assist , either ported or passive reflex or any of the other bass extension technologies. Even a three way system will still need the assistance of a real subwoofer by this time its more about power than anything else.
Due to the size and characteristics of our listening spaces, these designs aim for a directional response in the highs, dipole response in the mids and omni directional response in the lows.
So I asked chat gpt to design a speaker to see whether it is work ready and ready to work for us. How would you improve chat gpts design, do we need to add details about box stuffing or lining? Crossover points and slopes etc? Are the box dimensions right? Has speaker design plateaued or will new technologies arise?
The enclosure should be constructed from 3/4" MDF or plywood with internal bracing to minimize resonance and vibration. The enclosure walls should be finished with a vinyl wrap or painted finish to suit personal preferences.
The woofer chambers should each feature a 2" diameter x 7" long port, placed near the bottom of the enclosure, to tune the bass response for the DC200-8 woofers. The midrange chamber should be isolated from the woofers and designed to provide optimal performance for the DC160-8 driver. The tweeter chamber should be designed to provide optimal dispersion and smooth response for the ND25FA-4 tweeter.
An interesting video
Once a suitable 6 inch driver is chosen , a suitable one inch tweeter is added to the design. Munson showed that were more concerned with the 125Hz to 16kHz listening range which is effectively covered by a two way design through careful selection of a 6 inch driver and a 1 inch driver. Full rangers argue there's no content at 16kHz and choose a single 6 inch driver placed within a suitable enclosure that minimizes cone movement to control modulation. This summarizes speaker design.
Anything after 125Hz requires power, surface area, excursion. To be modest in two way just add another 6 inch driver, popular with Klipsch. With more amplifier power one can go three way by adding two 8 inch or two 10 inch drivers with bass reflex assist , either ported or passive reflex or any of the other bass extension technologies. Even a three way system will still need the assistance of a real subwoofer by this time its more about power than anything else.
Due to the size and characteristics of our listening spaces, these designs aim for a directional response in the highs, dipole response in the mids and omni directional response in the lows.
So I asked chat gpt to design a speaker to see whether it is work ready and ready to work for us. How would you improve chat gpts design, do we need to add details about box stuffing or lining? Crossover points and slopes etc? Are the box dimensions right? Has speaker design plateaued or will new technologies arise?
Design Overview
This speaker system is a three-way design featuring a 1" soft dome neodymium tweeter, a 6-1/2" midrange driver, and two 8" woofers. The enclosure is a bass reflex design with separate chambers for each driver, constructed from 3/4" MDF or plywood with internal bracing. The system is designed to provide a smooth and extended frequency response from 30 Hz to 20 kHz.Driver Selection
The following drivers have been selected for this system:- Tweeter: Dayton Audio ND25FA-4 1" Soft Dome Neodymium Tweeter
- Midrange: Dayton Audio DC160-8 6-1/2" Classic Woofer
- Woofers: Dayton Audio DC200-8 8" Classic Woofer (x2)
Crossover Design
The crossover design for this system consists of the following components:- High-pass filter for the tweeter at 2 kHz with a 12 dB/octave slope.
- Low-pass filter for the midrange at 300 Hz with a 12 dB/octave slope.
- High-pass filter for the woofers at 300 Hz with a 12 dB/octave slope.
Enclosure Design
The enclosure for this system is a three-way bass reflex design with separate chambers for each driver. The enclosure dimensions are 28" H x 16" W x 14" D (external dimensions), with the internal volume divided into three separate chambers using internal walls.The enclosure should be constructed from 3/4" MDF or plywood with internal bracing to minimize resonance and vibration. The enclosure walls should be finished with a vinyl wrap or painted finish to suit personal preferences.
The woofer chambers should each feature a 2" diameter x 7" long port, placed near the bottom of the enclosure, to tune the bass response for the DC200-8 woofers. The midrange chamber should be isolated from the woofers and designed to provide optimal performance for the DC160-8 driver. The tweeter chamber should be designed to provide optimal dispersion and smooth response for the ND25FA-4 tweeter.
Performance Specifications
The following performance specifications are expected for this speaker system:- Frequency response: 30 Hz - 20 kHz (+/- 3 dB)
- Sensitivity: 89 dB (1W/1m)
- Impedance: 8 ohms
- Power handling: 150 watts RMS, 300 watts peak
Conclusion
This speaker system design features high-quality and affordable drivers, a carefully tuned crossover, and a well-designed enclosure to provide a smooth and extended frequency response. The system is ideal for use in home audio applications and is designed to provide excellent performance and value for the price.An interesting video
Hi, isn't this the way speakers have been designed tens of years? You can improve the outcome by changing how and what you prompt from it, key thing.
In my opinion the process described in your post is backwards. Thats choosing some drivers and speaker system topology without any particular reason. Price point is valid reason but when you factor in issues like room and application, SPL capability and so forth it could be more expensive than necessary so even then it looks like no reason speaker to me.
You should try it again, you could start prompting from what is good perceived sound quality and how to reach it in a given context, with given application in mind. How room acoustics change the situation and so on. Depending on the answers continue prompting until you have enough understanding what the system could be like to make good perceived sound quality in a situation you are working with.
In the end you'd probably need to know something about speakers in general to be able to actually realize a system, debug it if there was misinformation in the chat process and so on 🙂
Example for initial prompt:
Then start asking additional questions form each bullet point and what ever is it that you don't know, better still ask for everything, question anything, take notes. Reflect to reality if its just non-sense. In the end, ask how to realize such speaker 🙂 Questions I would ask would be like, how room affects the frequency response, how much is enough SPL capability to undistorted dynamics, what makes clarity, and so on, what the position of speaker and listener should be in regards of typical furnishing for positive impact for sound quality and so on.
99% of a speaker design is missing from your initial post, selection of drivers is second last step. Last step is to measure the construct and simulate and build and tweak the xo which you can do nowadays pretty much perfectly with VituixCAD, both pretty much trivial tasks after you (or chatGPT) have designed the system, a target to aim for 😉
Have fun!
In my opinion the process described in your post is backwards. Thats choosing some drivers and speaker system topology without any particular reason. Price point is valid reason but when you factor in issues like room and application, SPL capability and so forth it could be more expensive than necessary so even then it looks like no reason speaker to me.
You should try it again, you could start prompting from what is good perceived sound quality and how to reach it in a given context, with given application in mind. How room acoustics change the situation and so on. Depending on the answers continue prompting until you have enough understanding what the system could be like to make good perceived sound quality in a situation you are working with.
In the end you'd probably need to know something about speakers in general to be able to actually realize a system, debug it if there was misinformation in the chat process and so on 🙂
Example for initial prompt:
Then start asking additional questions form each bullet point and what ever is it that you don't know, better still ask for everything, question anything, take notes. Reflect to reality if its just non-sense. In the end, ask how to realize such speaker 🙂 Questions I would ask would be like, how room affects the frequency response, how much is enough SPL capability to undistorted dynamics, what makes clarity, and so on, what the position of speaker and listener should be in regards of typical furnishing for positive impact for sound quality and so on.
99% of a speaker design is missing from your initial post, selection of drivers is second last step. Last step is to measure the construct and simulate and build and tweak the xo which you can do nowadays pretty much perfectly with VituixCAD, both pretty much trivial tasks after you (or chatGPT) have designed the system, a target to aim for 😉
Have fun!
Last edited:
Perhaps my approach to speaker design is no better than whats in the opening post. Perhaps there is better approach to interact with the AI!? Hopefully people try it out and share if they find interesting things. All we care is nice sound with the resources we have, enjoying music and having fun with a hobby, right? It would be nice if anyone could have that without years of study and fat wallet 🙂
Last edited:
Have you checked the research behind Munson's claim (assuming he made that claim) to make sure it is applicable?
For instance is it applicable in home situated sound reproduction?
For instance is it applicable in home situated sound reproduction?
There is probably million ways to approach these tools. Below is one, just regular conversation type of thing starting somewhere and then led to more specific stuff. Minus side here is that one has to know what to ask because the ChatGPT starts to point to a professional loudspeaker designer / acoustician and for example the math seems to have some issues 🙂 Pointing to a real professional is good advice as such. I would think currently that this kind of a long winded discussion would give much better rate of success how to make a good loudspeaker than just asking for blueprints. Well, perhaps it would give nice blueprints, never asked 😀
Me:
how to make very good loudspeakers?
ChatGPT:
Designing and building high-quality loudspeakers requires a combination of technical expertise, attention to detail, and a good understanding of the principles of acoustics. Here are some general steps that can help you create very good loudspeakers:
Determine the application: What will be the primary use of the loudspeaker? Is it for home audio, professional audio, or something else? This will help you decide on the size and configuration of the speakers, as well as the components and materials you will need.
Choose the drivers: The drivers are the components that actually produce the sound in the speaker. This includes the woofer for low frequencies, the tweeter for high frequencies, and possibly a mid-range driver. Choose drivers that have a good frequency response, low distortion, and a high power handling capability.
Design the cabinet: The cabinet is the enclosure that houses the drivers. The design of the cabinet can have a significant impact on the sound quality of the loudspeaker. Factors to consider include the size and shape of the cabinet, the materials used, and the placement and configuration of the drivers.
Select the crossover: The crossover is the component that divides the audio signal into different frequency bands and sends them to the appropriate driver. A good crossover will ensure that each driver receives the appropriate frequencies and that the sound is balanced and coherent.
Assemble and test: Once you have all the components, assemble the loudspeakers and test them thoroughly. This includes measuring the frequency response, checking for distortion, and listening for any anomalies or issues.
Fine-tune and adjust: Finally, fine-tune the speakers by adjusting the crossover, cabinet design, and other factors as necessary to achieve the desired sound quality.
It's important to note that designing and building high-quality loudspeakers can be a complex and challenging task, and may require specialized tools and expertise. If you're not confident in your ability to design and build speakers from scratch, consider using a pre-designed kit or working with a professional speaker designer or manufacturer.
Me:
Application is music listening in living room. Its an apartment so cannot listen very loud. Whats loud by the way?
ChatGPT:
In the context of a living room in an apartment, "loud" would typically refer to a sound level that is sufficient to fill the space and provide an enjoyable listening experience, but not so loud as to disturb neighbors or cause damage to hearing.
The actual loudness level that is considered "loud" can vary depending on individual preferences, room size and layout, and the quality of the speakers. In general, a sound level of around 80-85 decibels (dB) is considered reasonable for music listening in a residential setting. However, this can vary depending on the specific circumstances, and it's always a good idea to be considerate of neighbors and others who may be sharing the space.
Me:
Is this mean value or peak value?
ChatGPT:
The sound level values typically referenced in audio and acoustics are usually measured in decibels (dB) SPL (Sound Pressure Level), which is a measure of the sound pressure level relative to a reference pressure level of 20 micropascals (μPa).
In most cases, when people talk about a specific sound level, they are referring to the A-weighted sound pressure level (dBA), which is a measure of sound pressure levels that are adjusted to correspond more closely to the way the human ear perceives sound.
The dBA scale is typically based on a time-weighted average (TWA) of the sound pressure levels over a given period of time, usually in one-second intervals. This means that the value represents an average sound pressure level over that time period, rather than a peak or instantaneous value.
That being said, it's important to note that the actual sound pressure level (SPL) of music or any other sound can vary significantly depending on the specific recording, playback equipment, listening environment, and other factors. The peak values of a sound can be much higher than the mean or average values, and it's important to be aware of the potential for hearing damage at higher SPLs.
Me:
How much higher peaks can be? Thinking of the loudspeaker design that would be capable of 80db average at listening spot, how much its SPL capability should be to properly reproduce Jazz records for example?
ChatGPT:
The peak sound pressure levels (SPL) of music can vary greatly depending on the recording, instrumentation, and other factors. In general, music can have peak SPLs that are 10-20 dB higher than the average or mean SPL, and in some cases even higher.
For example, a well-recorded and dynamically compressed jazz album may have an average SPL of 80 dB at the listening spot, but could have peak levels of 90-100 dB or even higher during the loudest sections.
To properly reproduce this type of music, a loudspeaker system should be capable of handling these peak levels without distortion or damage. As a general rule of thumb, the peak SPL capability of a loudspeaker system should be at least 10-20 dB higher than the expected average SPL at the listening position.
For a system that can deliver an average SPL of 80 dB at the listening position, a peak SPL capability of 90-100 dB or higher would be appropriate to ensure proper reproduction of jazz music and other dynamic recordings. It's important to note that the specific requirements will depend on the specific listening environment, the preferences of the listener, and other factors.
Me:
What kind of a system we are talking about then? If we need 100db SPL capability for bandwidth from 30Hz to 20kHz what kind of loudspeaker system would be capable of it?
ChatGPT
Achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz would require a high-quality loudspeaker system with a powerful amplifier and carefully designed components.
The specific type of loudspeaker system required will depend on a variety of factors, including the size and shape of the listening room, the desired listening distance, the type of music being played, and other factors.
In general, to achieve this level of SPL capability, a loudspeaker system would need to have the following characteristics:
High-efficiency drivers: The drivers (woofers, tweeters, etc.) should be designed to provide a high level of sound output for a given input power. This can be achieved through the use of lightweight diaphragms, powerful magnets, and other design features.
Large amplifiers: To drive the loudspeaker system to high levels, a large, high-quality amplifier with a high current output capability will be required. This will ensure that the system can deliver the necessary power to the drivers without distortion or other issues.
Carefully designed crossover: The crossover circuitry that separates the audio signal into different frequency bands and directs them to the appropriate drivers will need to be carefully designed and tuned to ensure proper frequency response and phase alignment.
Carefully designed enclosure: The loudspeaker enclosure should be designed to minimize resonances and other sources of distortion, and to provide a high level of efficiency.
Proper room placement: The loudspeaker system should be placed in a position that maximizes its efficiency and minimizes unwanted reflections and other acoustic issues in the room.
Overall, achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz is a significant technical challenge, and would require a high-quality, well-designed loudspeaker system with carefully matched components and proper installation and setup.
Me:
What kind of components are we talking about? Would 6.5" woofer and 1" tweeter two way speaker be capable of it? How much power that would require if it is?
ChatGPT:
Achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz with a 6.5" woofer and 1" tweeter two-way speaker would be very challenging, if not impossible.
In general, larger drivers and more powerful amplifiers are needed to achieve high sound pressure levels, particularly at lower frequencies. While a 6.5" woofer can provide good midrange and upper bass response, it would likely struggle to reproduce lower bass frequencies at high levels without distortion or damage. Similarly, a 1" tweeter may not be able to provide the necessary high-frequency output at high levels without distortion.
To achieve the required SPL levels, a loudspeaker system with larger drivers (e.g. 8" or larger) and a more powerful amplifier would likely be necessary. In addition, a well-designed enclosure with proper acoustic treatment and damping materials would be needed to minimize resonances and other sources of distortion.
The power required to achieve a peak SPL of 100 dB will depend on the specific loudspeaker system and listening environment. However, in general, a power amplifier with a minimum output of around 100 watts RMS per channel would be necessary to achieve this level of SPL with a loudspeaker system capable of handling it. The actual power required will depend on factors such as the sensitivity and impedance of the loudspeaker system, the size and acoustics of the listening room, and other factors.
Me:
How to calculate what kind of woofer is required to reproduce 30Hz at 100db in a living room?
ChatGPT:
To calculate what kind of woofer is required to reproduce 30 Hz at 100 dB in a living room, you will need to consider a few factors:
Desired sound pressure level: 100 dB at 30 Hz is a very high sound pressure level, and would require a high-output, high-power woofer to achieve.
Room size and acoustics: The size and acoustics of the room will play a significant role in determining how much power is required to achieve the desired sound pressure level. Larger rooms with hard surfaces will require more power to achieve the same SPL as smaller, more acoustically treated rooms.
Loudspeaker efficiency: The efficiency of the loudspeaker will determine how much power is required to achieve the desired SPL. A more efficient loudspeaker will require less power to achieve the same SPL as a less efficient one.
Low-frequency extension: In order to reproduce 30 Hz at high levels, a loudspeaker with a low-frequency extension below 30 Hz will likely be necessary. This will require a woofer with a large diameter and a well-designed enclosure.
Given these factors, it's difficult to provide a precise calculation for the kind of woofer required to achieve 100 dB at 30 Hz in a living room. However, in general, a high-quality 10-12 inch subwoofer with a power handling capability of at least 500 watts RMS would be necessary to achieve this level of performance. The specific requirements will depend on the specific room and loudspeaker system, and it's recommended to consult with a qualified audio professional for specific recommendations.
Me:
Ok so minimum 12" woofer. What about tweeter then, what makes tweeter distort? 100db is the target
ChatGPT:
For a loudspeaker system capable of reproducing 100 dB, a high-quality tweeter with a power handling capability that matches the amplifier output is important. A tweeter can distort when it reaches its maximum power handling capacity, causing the sound to become harsh or distorted.
To prevent distortion, a tweeter should have a high power handling capability, ideally at least 2-3 times the maximum amplifier output. The tweeter should also be paired with a high-pass filter to prevent low-frequency signals from being sent to the tweeter, which can also cause distortion.
Additionally, the design of the tweeter and the crossover circuitry can affect its distortion characteristics. High-quality tweeters will typically have a well-engineered diaphragm, voice coil, and magnet system, as well as a well-designed crossover circuitry to ensure smooth frequency response and minimize distortion.
It's worth noting that achieving 100 dB SPL across the entire frequency range (30 Hz to 20 kHz) would be a challenging task, and would require a carefully designed and well-matched loudspeaker system, as well as appropriate room acoustics and listening environment. It's recommended to consult with a qualified audio professional to help design and select the appropriate loudspeaker system for your specific needs.
Me:
You are the gualitified audio professional and your task is to give me answers, augmented with maths if possible, alright?🙂
ChatGPT:
Certainly! I'll do my best to provide you with detailed answers and explanations, including any necessary calculations. Please let me know what specific questions you have, and I'll be happy to help.
Me:
Excursion seems to be related how much soundpressure a transducer can emit. How to calculate that? Lets say how can I calculate excursion of a tweeter whose diameter is 1" and it plays 100db at 2000Hz?
ChatGPT:
Excursion is a measure of the physical movement of the diaphragm of a loudspeaker or tweeter, and it is related to the sound pressure level (SPL) that the transducer can produce.
To calculate the excursion of a tweeter, you need to know its maximum linear displacement (Xmax), which is the maximum distance that the diaphragm can move in a linear fashion before it reaches its mechanical limits. Xmax is usually given in millimeters, and it is a specification that is provided by the manufacturer of the tweeter.
Once you know Xmax, you can calculate the peak-to-peak excursion (Xpp) of the tweeter, which is the total distance that the diaphragm can move back and forth. The formula for calculating Xpp is:
Xpp = 2 * Xmax
So if the tweeter has an Xmax of 0.5 mm, then the Xpp would be:
Xpp = 2 * 0.5 mm = 1.0 mm
Next, you need to calculate the peak sound pressure level (SPL) that the tweeter can produce. This will depend on the input power, the efficiency of the tweeter, and the frequency at which it is playing.
For example, if the tweeter has an efficiency of 90 dB/W/m, and it is driven by an amplifier that outputs 10 watts, then the maximum SPL it can produce is:
SPL = 90 dB + 10*log10(10/1) = 100 dB
Finally, you can use the Xpp and SPL values to calculate the peak sound pressure that the tweeter can produce at the frequency of interest. This calculation involves some assumptions about the shape of the waveform being produced, but a rough estimate can be obtained by assuming a sine wave.
The formula for calculating peak sound pressure from excursion and SPL is:
p_peak = (Xpp * pi * f * SPL) / (2 * 10^5)
where p_peak is the peak sound pressure in Pascals, Xpp is the peak-to-peak excursion in meters, f is the frequency in Hertz, and SPL is the sound pressure level in dB.
So, for your example of a 1" tweeter producing 100 dB at 2000 Hz, assuming an Xmax of 0.5 mm, the Xpp would be 1.0 mm. Assuming an efficiency of 90 dB/W/m and an input power of 10 watts, the maximum SPL would be 100 dB. Using the above formula, the peak sound pressure at 2000 Hz would be:
p_peak = (1.0 mm * pi * 2000 Hz * 100 dB) / (2 * 10^5) = 0.31 Pa
Keep in mind that this is a rough estimate, and actual performance will depend on many factors, including the design of the tweeter, the acoustic environment, and the specific waveform being produced.
Me:
How about woofers xmax. We have 12" driver in a closed box, what would excursion be at 30Hz when SPL is 100db?
ChatGPT:
Calculating the required excursion (Xmax) of a 12" woofer to produce 100 dB SPL at 30 Hz in a closed box is a bit more involved, but we can estimate it using some simplifying assumptions.
First, we need to calculate the acoustic power required to produce 100 dB SPL at 30 Hz in a typical living room environment. Assuming a typical living room size of around 20 square meters, and assuming a diffuse sound field, we can estimate the required acoustic power as follows:
P_acoustic = 10^(100/10) * (4piR^2) / (4pir^2) * (1/16) * (1/10)
where P_acoustic is the required acoustic power in watts, R is the listening distance in meters (let's assume 3 meters), r is the room dimension in meters (let's assume 5 meters), 1/16 is the assumed reverberation time of the room in seconds, and 1/10 is the assumed absorption coefficient of the room.
Plugging in the numbers, we get:
P_acoustic = 10^(100/10) * (4pi3^2) / (4pi5^2) * (1/16) * (1/10) = 0.007 watts
Next, we need to calculate the cone displacement required to produce this acoustic power at 30 Hz. This can be done using the formula for sound power radiated by a loudspeaker:
P_acoustic = (2pirhoc^2A^2f^3Xmax^2) / (3*r^2)
where rho is the air density (1.2 kg/m^3), c is the speed of sound (343 m/s), A is the area of the cone (pi*r^2), f is the frequency in Hz, and Xmax is the required peak-to-peak cone displacement in meters.
Solving for Xmax, we get:
Xmax = sqrt((3P_acousticr^2) / (2pirhoc^2A^2*f^3))
Assuming a 12" woofer with a cone area of around 0.05 square meters (pi*(12*0.0254/2)^2), we get:
Xmax = sqrt((30.0073^2) / (2pi1.2343^20.05^2*30^3)) = 1.1 mm
So, a 12" woofer with an Xmax of around 1.1 mm should be able to produce 100 dB SPL at 30 Hz in a closed box in a typical living room environment. Keep in mind that this is a rough estimate, and actual performance will depend on many factors, including the design of the woofer, the acoustic environment, and the specific waveform being produced.
.....
how to make very good loudspeakers?
ChatGPT:
Designing and building high-quality loudspeakers requires a combination of technical expertise, attention to detail, and a good understanding of the principles of acoustics. Here are some general steps that can help you create very good loudspeakers:
Determine the application: What will be the primary use of the loudspeaker? Is it for home audio, professional audio, or something else? This will help you decide on the size and configuration of the speakers, as well as the components and materials you will need.
Choose the drivers: The drivers are the components that actually produce the sound in the speaker. This includes the woofer for low frequencies, the tweeter for high frequencies, and possibly a mid-range driver. Choose drivers that have a good frequency response, low distortion, and a high power handling capability.
Design the cabinet: The cabinet is the enclosure that houses the drivers. The design of the cabinet can have a significant impact on the sound quality of the loudspeaker. Factors to consider include the size and shape of the cabinet, the materials used, and the placement and configuration of the drivers.
Select the crossover: The crossover is the component that divides the audio signal into different frequency bands and sends them to the appropriate driver. A good crossover will ensure that each driver receives the appropriate frequencies and that the sound is balanced and coherent.
Assemble and test: Once you have all the components, assemble the loudspeakers and test them thoroughly. This includes measuring the frequency response, checking for distortion, and listening for any anomalies or issues.
Fine-tune and adjust: Finally, fine-tune the speakers by adjusting the crossover, cabinet design, and other factors as necessary to achieve the desired sound quality.
It's important to note that designing and building high-quality loudspeakers can be a complex and challenging task, and may require specialized tools and expertise. If you're not confident in your ability to design and build speakers from scratch, consider using a pre-designed kit or working with a professional speaker designer or manufacturer.
Me:
Application is music listening in living room. Its an apartment so cannot listen very loud. Whats loud by the way?
ChatGPT:
In the context of a living room in an apartment, "loud" would typically refer to a sound level that is sufficient to fill the space and provide an enjoyable listening experience, but not so loud as to disturb neighbors or cause damage to hearing.
The actual loudness level that is considered "loud" can vary depending on individual preferences, room size and layout, and the quality of the speakers. In general, a sound level of around 80-85 decibels (dB) is considered reasonable for music listening in a residential setting. However, this can vary depending on the specific circumstances, and it's always a good idea to be considerate of neighbors and others who may be sharing the space.
Me:
Is this mean value or peak value?
ChatGPT:
The sound level values typically referenced in audio and acoustics are usually measured in decibels (dB) SPL (Sound Pressure Level), which is a measure of the sound pressure level relative to a reference pressure level of 20 micropascals (μPa).
In most cases, when people talk about a specific sound level, they are referring to the A-weighted sound pressure level (dBA), which is a measure of sound pressure levels that are adjusted to correspond more closely to the way the human ear perceives sound.
The dBA scale is typically based on a time-weighted average (TWA) of the sound pressure levels over a given period of time, usually in one-second intervals. This means that the value represents an average sound pressure level over that time period, rather than a peak or instantaneous value.
That being said, it's important to note that the actual sound pressure level (SPL) of music or any other sound can vary significantly depending on the specific recording, playback equipment, listening environment, and other factors. The peak values of a sound can be much higher than the mean or average values, and it's important to be aware of the potential for hearing damage at higher SPLs.
Me:
How much higher peaks can be? Thinking of the loudspeaker design that would be capable of 80db average at listening spot, how much its SPL capability should be to properly reproduce Jazz records for example?
ChatGPT:
The peak sound pressure levels (SPL) of music can vary greatly depending on the recording, instrumentation, and other factors. In general, music can have peak SPLs that are 10-20 dB higher than the average or mean SPL, and in some cases even higher.
For example, a well-recorded and dynamically compressed jazz album may have an average SPL of 80 dB at the listening spot, but could have peak levels of 90-100 dB or even higher during the loudest sections.
To properly reproduce this type of music, a loudspeaker system should be capable of handling these peak levels without distortion or damage. As a general rule of thumb, the peak SPL capability of a loudspeaker system should be at least 10-20 dB higher than the expected average SPL at the listening position.
For a system that can deliver an average SPL of 80 dB at the listening position, a peak SPL capability of 90-100 dB or higher would be appropriate to ensure proper reproduction of jazz music and other dynamic recordings. It's important to note that the specific requirements will depend on the specific listening environment, the preferences of the listener, and other factors.
Me:
What kind of a system we are talking about then? If we need 100db SPL capability for bandwidth from 30Hz to 20kHz what kind of loudspeaker system would be capable of it?
ChatGPT
Achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz would require a high-quality loudspeaker system with a powerful amplifier and carefully designed components.
The specific type of loudspeaker system required will depend on a variety of factors, including the size and shape of the listening room, the desired listening distance, the type of music being played, and other factors.
In general, to achieve this level of SPL capability, a loudspeaker system would need to have the following characteristics:
High-efficiency drivers: The drivers (woofers, tweeters, etc.) should be designed to provide a high level of sound output for a given input power. This can be achieved through the use of lightweight diaphragms, powerful magnets, and other design features.
Large amplifiers: To drive the loudspeaker system to high levels, a large, high-quality amplifier with a high current output capability will be required. This will ensure that the system can deliver the necessary power to the drivers without distortion or other issues.
Carefully designed crossover: The crossover circuitry that separates the audio signal into different frequency bands and directs them to the appropriate drivers will need to be carefully designed and tuned to ensure proper frequency response and phase alignment.
Carefully designed enclosure: The loudspeaker enclosure should be designed to minimize resonances and other sources of distortion, and to provide a high level of efficiency.
Proper room placement: The loudspeaker system should be placed in a position that maximizes its efficiency and minimizes unwanted reflections and other acoustic issues in the room.
Overall, achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz is a significant technical challenge, and would require a high-quality, well-designed loudspeaker system with carefully matched components and proper installation and setup.
Me:
What kind of components are we talking about? Would 6.5" woofer and 1" tweeter two way speaker be capable of it? How much power that would require if it is?
ChatGPT:
Achieving a peak SPL capability of 100 dB over a bandwidth from 30 Hz to 20 kHz with a 6.5" woofer and 1" tweeter two-way speaker would be very challenging, if not impossible.
In general, larger drivers and more powerful amplifiers are needed to achieve high sound pressure levels, particularly at lower frequencies. While a 6.5" woofer can provide good midrange and upper bass response, it would likely struggle to reproduce lower bass frequencies at high levels without distortion or damage. Similarly, a 1" tweeter may not be able to provide the necessary high-frequency output at high levels without distortion.
To achieve the required SPL levels, a loudspeaker system with larger drivers (e.g. 8" or larger) and a more powerful amplifier would likely be necessary. In addition, a well-designed enclosure with proper acoustic treatment and damping materials would be needed to minimize resonances and other sources of distortion.
The power required to achieve a peak SPL of 100 dB will depend on the specific loudspeaker system and listening environment. However, in general, a power amplifier with a minimum output of around 100 watts RMS per channel would be necessary to achieve this level of SPL with a loudspeaker system capable of handling it. The actual power required will depend on factors such as the sensitivity and impedance of the loudspeaker system, the size and acoustics of the listening room, and other factors.
Me:
How to calculate what kind of woofer is required to reproduce 30Hz at 100db in a living room?
ChatGPT:
To calculate what kind of woofer is required to reproduce 30 Hz at 100 dB in a living room, you will need to consider a few factors:
Desired sound pressure level: 100 dB at 30 Hz is a very high sound pressure level, and would require a high-output, high-power woofer to achieve.
Room size and acoustics: The size and acoustics of the room will play a significant role in determining how much power is required to achieve the desired sound pressure level. Larger rooms with hard surfaces will require more power to achieve the same SPL as smaller, more acoustically treated rooms.
Loudspeaker efficiency: The efficiency of the loudspeaker will determine how much power is required to achieve the desired SPL. A more efficient loudspeaker will require less power to achieve the same SPL as a less efficient one.
Low-frequency extension: In order to reproduce 30 Hz at high levels, a loudspeaker with a low-frequency extension below 30 Hz will likely be necessary. This will require a woofer with a large diameter and a well-designed enclosure.
Given these factors, it's difficult to provide a precise calculation for the kind of woofer required to achieve 100 dB at 30 Hz in a living room. However, in general, a high-quality 10-12 inch subwoofer with a power handling capability of at least 500 watts RMS would be necessary to achieve this level of performance. The specific requirements will depend on the specific room and loudspeaker system, and it's recommended to consult with a qualified audio professional for specific recommendations.
Me:
Ok so minimum 12" woofer. What about tweeter then, what makes tweeter distort? 100db is the target
ChatGPT:
For a loudspeaker system capable of reproducing 100 dB, a high-quality tweeter with a power handling capability that matches the amplifier output is important. A tweeter can distort when it reaches its maximum power handling capacity, causing the sound to become harsh or distorted.
To prevent distortion, a tweeter should have a high power handling capability, ideally at least 2-3 times the maximum amplifier output. The tweeter should also be paired with a high-pass filter to prevent low-frequency signals from being sent to the tweeter, which can also cause distortion.
Additionally, the design of the tweeter and the crossover circuitry can affect its distortion characteristics. High-quality tweeters will typically have a well-engineered diaphragm, voice coil, and magnet system, as well as a well-designed crossover circuitry to ensure smooth frequency response and minimize distortion.
It's worth noting that achieving 100 dB SPL across the entire frequency range (30 Hz to 20 kHz) would be a challenging task, and would require a carefully designed and well-matched loudspeaker system, as well as appropriate room acoustics and listening environment. It's recommended to consult with a qualified audio professional to help design and select the appropriate loudspeaker system for your specific needs.
Me:
You are the gualitified audio professional and your task is to give me answers, augmented with maths if possible, alright?🙂
ChatGPT:
Certainly! I'll do my best to provide you with detailed answers and explanations, including any necessary calculations. Please let me know what specific questions you have, and I'll be happy to help.
Me:
Excursion seems to be related how much soundpressure a transducer can emit. How to calculate that? Lets say how can I calculate excursion of a tweeter whose diameter is 1" and it plays 100db at 2000Hz?
ChatGPT:
Excursion is a measure of the physical movement of the diaphragm of a loudspeaker or tweeter, and it is related to the sound pressure level (SPL) that the transducer can produce.
To calculate the excursion of a tweeter, you need to know its maximum linear displacement (Xmax), which is the maximum distance that the diaphragm can move in a linear fashion before it reaches its mechanical limits. Xmax is usually given in millimeters, and it is a specification that is provided by the manufacturer of the tweeter.
Once you know Xmax, you can calculate the peak-to-peak excursion (Xpp) of the tweeter, which is the total distance that the diaphragm can move back and forth. The formula for calculating Xpp is:
Xpp = 2 * Xmax
So if the tweeter has an Xmax of 0.5 mm, then the Xpp would be:
Xpp = 2 * 0.5 mm = 1.0 mm
Next, you need to calculate the peak sound pressure level (SPL) that the tweeter can produce. This will depend on the input power, the efficiency of the tweeter, and the frequency at which it is playing.
For example, if the tweeter has an efficiency of 90 dB/W/m, and it is driven by an amplifier that outputs 10 watts, then the maximum SPL it can produce is:
SPL = 90 dB + 10*log10(10/1) = 100 dB
Finally, you can use the Xpp and SPL values to calculate the peak sound pressure that the tweeter can produce at the frequency of interest. This calculation involves some assumptions about the shape of the waveform being produced, but a rough estimate can be obtained by assuming a sine wave.
The formula for calculating peak sound pressure from excursion and SPL is:
p_peak = (Xpp * pi * f * SPL) / (2 * 10^5)
where p_peak is the peak sound pressure in Pascals, Xpp is the peak-to-peak excursion in meters, f is the frequency in Hertz, and SPL is the sound pressure level in dB.
So, for your example of a 1" tweeter producing 100 dB at 2000 Hz, assuming an Xmax of 0.5 mm, the Xpp would be 1.0 mm. Assuming an efficiency of 90 dB/W/m and an input power of 10 watts, the maximum SPL would be 100 dB. Using the above formula, the peak sound pressure at 2000 Hz would be:
p_peak = (1.0 mm * pi * 2000 Hz * 100 dB) / (2 * 10^5) = 0.31 Pa
Keep in mind that this is a rough estimate, and actual performance will depend on many factors, including the design of the tweeter, the acoustic environment, and the specific waveform being produced.
Me:
How about woofers xmax. We have 12" driver in a closed box, what would excursion be at 30Hz when SPL is 100db?
ChatGPT:
Calculating the required excursion (Xmax) of a 12" woofer to produce 100 dB SPL at 30 Hz in a closed box is a bit more involved, but we can estimate it using some simplifying assumptions.
First, we need to calculate the acoustic power required to produce 100 dB SPL at 30 Hz in a typical living room environment. Assuming a typical living room size of around 20 square meters, and assuming a diffuse sound field, we can estimate the required acoustic power as follows:
P_acoustic = 10^(100/10) * (4piR^2) / (4pir^2) * (1/16) * (1/10)
where P_acoustic is the required acoustic power in watts, R is the listening distance in meters (let's assume 3 meters), r is the room dimension in meters (let's assume 5 meters), 1/16 is the assumed reverberation time of the room in seconds, and 1/10 is the assumed absorption coefficient of the room.
Plugging in the numbers, we get:
P_acoustic = 10^(100/10) * (4pi3^2) / (4pi5^2) * (1/16) * (1/10) = 0.007 watts
Next, we need to calculate the cone displacement required to produce this acoustic power at 30 Hz. This can be done using the formula for sound power radiated by a loudspeaker:
P_acoustic = (2pirhoc^2A^2f^3Xmax^2) / (3*r^2)
where rho is the air density (1.2 kg/m^3), c is the speed of sound (343 m/s), A is the area of the cone (pi*r^2), f is the frequency in Hz, and Xmax is the required peak-to-peak cone displacement in meters.
Solving for Xmax, we get:
Xmax = sqrt((3P_acousticr^2) / (2pirhoc^2A^2*f^3))
Assuming a 12" woofer with a cone area of around 0.05 square meters (pi*(12*0.0254/2)^2), we get:
Xmax = sqrt((30.0073^2) / (2pi1.2343^20.05^2*30^3)) = 1.1 mm
So, a 12" woofer with an Xmax of around 1.1 mm should be able to produce 100 dB SPL at 30 Hz in a closed box in a typical living room environment. Keep in mind that this is a rough estimate, and actual performance will depend on many factors, including the design of the woofer, the acoustic environment, and the specific waveform being produced.
.....
Last edited:
Trying again and it seems that loudspeaker stuff is quite niche, answers seem to evolve to some of the most popular loudspeaker systems like two way bookshelf speaker. While such speaker might or might not be the pinnacle for home stereo certainly getting anything else from chatGPT seems to be tough and require knowledge in the prompts to lead to another direction, to more specific stuff. It seems to be too general model to get anything above average.
Starts of conversations seem nice, determine goals/application first, but there doesn't seem to be that much information behind after few prompts. Perhaps I'm just a fool and asking wrong things. Is it arrogance to think I know better and that there would be audio quality improvement past two way bookshelf speaker?🙂 It doesn't seem to give output I'm looking for, instead some answers seem to be plain wrong (to me) and so on... but I still think its useful tool. I think its good to chat with it because it encourages me to think the stuff, to raise new questions which can give insight and question old habits of mine, question the stuff I think I knew, question answers chatgpt gives and so on, if nothing else. Encourages thinking and gives experience communicating with such system if nothing else.
Anyone have access to all AES papers and have hardware to train more specialized language model for loudspeaker / audio / heading / acoustics domain?🙂
Starts of conversations seem nice, determine goals/application first, but there doesn't seem to be that much information behind after few prompts. Perhaps I'm just a fool and asking wrong things. Is it arrogance to think I know better and that there would be audio quality improvement past two way bookshelf speaker?🙂 It doesn't seem to give output I'm looking for, instead some answers seem to be plain wrong (to me) and so on... but I still think its useful tool. I think its good to chat with it because it encourages me to think the stuff, to raise new questions which can give insight and question old habits of mine, question the stuff I think I knew, question answers chatgpt gives and so on, if nothing else. Encourages thinking and gives experience communicating with such system if nothing else.
Anyone have access to all AES papers and have hardware to train more specialized language model for loudspeaker / audio / heading / acoustics domain?🙂
Me:
How to approach loudspeaker design? Should I select drivers first or something else?
ChatGPT:
Designing a loudspeaker can be a complex process, but it generally involves several key steps. While there are different approaches to loudspeaker design, here's a general outline that you may find helpful:
Determine the desired application: Before starting any design work, it's important to understand the intended use of the loudspeaker. Will it be used for home theater, music playback, or something else? The answer to this question will affect many of the design choices you make.
Choose a basic design: There are several common loudspeaker designs, such as sealed, ported, and bandpass enclosures. Each has its own benefits and drawbacks, so it's important to choose the one that best fits your application.
Select drivers: Once you have chosen a basic design, you can begin selecting drivers. This step involves choosing the right size, type, and number of drivers to achieve the desired frequency response and power handling.
Design the enclosure: After selecting drivers, you will need to design the enclosure that will hold them. This involves determining the size and shape of the enclosure, as well as any additional features such as ports or passive radiators.
Fine-tune the design: With the basic design in place, you can start fine-tuning the loudspeaker to achieve the desired performance. This may involve adjusting the crossover network, selecting suitable damping materials, and optimizing the driver placement within the enclosure.
Overall, while selecting drivers is an important step in loudspeaker design, it's generally best to start by determining the intended application and choosing a basic design. From there, you can select drivers and make other design choices based on your goals and requirements.
Me:
how to determine which loudspeaker design suits my application? How to outline my application?
ChatGPT:
Determining which loudspeaker design suits your application depends on a variety of factors. Here are some steps you can take to outline your application and choose the right loudspeaker design:
Determine the intended use: Are you designing a loudspeaker for home theater, music playback, live sound, or some other application? The intended use will affect many of the design choices you make.
Determine the room size and acoustics: The size and shape of the room where the loudspeaker will be used will affect the overall sound quality. You'll need to consider factors such as room reflections, resonances, and other acoustic properties when choosing a design.
Determine the power requirements: How much power will the loudspeaker need to handle? This will depend on factors such as the size of the room, the intended listening level, and the type of music or content being played.
Determine the desired frequency response: Different loudspeaker designs have different frequency responses. You'll need to consider factors such as the desired bass extension, midrange clarity, and high-frequency detail when choosing a design.
Determine the budget: Loudspeaker design can be expensive, so you'll need to consider your budget when choosing a design. Keep in mind that higher-end designs may require more expensive components and materials.
Once you have outlined your application using these factors, you can begin researching and comparing different loudspeaker designs to find the one that best suits your needs. Consider factors such as the design's frequency response, power handling, and cost, and be sure to test out different designs if possible to find the one that sounds best in your particular application.
Me:
Its living room use, for music playback. I don't know what my room acoustics are, its a typical apartment living room. Big window, some furniture like sofas, hard walls, floor and ceiling. I cannot use much power, its expensive. I need to use small class-D amplifier that drinks small amount of power. Frequency response needs to be from 30Hz to 20kHz. Budget is limitless, manufacturing capability is also limitless but I don't want to use any audiophile nonsense, just good engineering principles.
ChatGPT:
Based on your requirements, here are some loudspeaker design recommendations that may suit your application:
Bookshelf loudspeakers: Bookshelf loudspeakers are a compact and versatile option for music playback in a living room. They are typically designed to be used on a bookshelf or stand and can be paired with a small class-D amplifier to keep power consumption low.
Sealed enclosure design: A sealed enclosure design is a simple and reliable option for a bookshelf loudspeaker. It provides good low-frequency extension and can be designed to have a flat frequency response across the desired frequency range.
Two-way design: A two-way design with a woofer and tweeter is a common choice for bookshelf loudspeakers. It can provide good midrange and high-frequency performance, and the woofer can be chosen to provide good low-frequency response within the constraints of a sealed enclosure.
Use well-regarded drivers: While you don't want to use audiophile nonsense, it's still important to choose high-quality drivers for your loudspeaker. Look for well-regarded drivers that have a reputation for good performance at reasonable prices.
Consider crossover design: The crossover network, which divides the signal between the woofer and tweeter, is an important part of loudspeaker design. A well-designed crossover can help ensure a flat frequency response and good performance across the desired frequency range.
Overall, a sealed two-way bookshelf loudspeaker design with well-regarded drivers and a carefully designed crossover should be able to meet your requirements for a living room music playback system. Be sure to prototype and test your design before committing to manufacturing to ensure that it performs as desired.
....
How to approach loudspeaker design? Should I select drivers first or something else?
ChatGPT:
Designing a loudspeaker can be a complex process, but it generally involves several key steps. While there are different approaches to loudspeaker design, here's a general outline that you may find helpful:
Determine the desired application: Before starting any design work, it's important to understand the intended use of the loudspeaker. Will it be used for home theater, music playback, or something else? The answer to this question will affect many of the design choices you make.
Choose a basic design: There are several common loudspeaker designs, such as sealed, ported, and bandpass enclosures. Each has its own benefits and drawbacks, so it's important to choose the one that best fits your application.
Select drivers: Once you have chosen a basic design, you can begin selecting drivers. This step involves choosing the right size, type, and number of drivers to achieve the desired frequency response and power handling.
Design the enclosure: After selecting drivers, you will need to design the enclosure that will hold them. This involves determining the size and shape of the enclosure, as well as any additional features such as ports or passive radiators.
Fine-tune the design: With the basic design in place, you can start fine-tuning the loudspeaker to achieve the desired performance. This may involve adjusting the crossover network, selecting suitable damping materials, and optimizing the driver placement within the enclosure.
Overall, while selecting drivers is an important step in loudspeaker design, it's generally best to start by determining the intended application and choosing a basic design. From there, you can select drivers and make other design choices based on your goals and requirements.
Me:
how to determine which loudspeaker design suits my application? How to outline my application?
ChatGPT:
Determining which loudspeaker design suits your application depends on a variety of factors. Here are some steps you can take to outline your application and choose the right loudspeaker design:
Determine the intended use: Are you designing a loudspeaker for home theater, music playback, live sound, or some other application? The intended use will affect many of the design choices you make.
Determine the room size and acoustics: The size and shape of the room where the loudspeaker will be used will affect the overall sound quality. You'll need to consider factors such as room reflections, resonances, and other acoustic properties when choosing a design.
Determine the power requirements: How much power will the loudspeaker need to handle? This will depend on factors such as the size of the room, the intended listening level, and the type of music or content being played.
Determine the desired frequency response: Different loudspeaker designs have different frequency responses. You'll need to consider factors such as the desired bass extension, midrange clarity, and high-frequency detail when choosing a design.
Determine the budget: Loudspeaker design can be expensive, so you'll need to consider your budget when choosing a design. Keep in mind that higher-end designs may require more expensive components and materials.
Once you have outlined your application using these factors, you can begin researching and comparing different loudspeaker designs to find the one that best suits your needs. Consider factors such as the design's frequency response, power handling, and cost, and be sure to test out different designs if possible to find the one that sounds best in your particular application.
Me:
Its living room use, for music playback. I don't know what my room acoustics are, its a typical apartment living room. Big window, some furniture like sofas, hard walls, floor and ceiling. I cannot use much power, its expensive. I need to use small class-D amplifier that drinks small amount of power. Frequency response needs to be from 30Hz to 20kHz. Budget is limitless, manufacturing capability is also limitless but I don't want to use any audiophile nonsense, just good engineering principles.
ChatGPT:
Based on your requirements, here are some loudspeaker design recommendations that may suit your application:
Bookshelf loudspeakers: Bookshelf loudspeakers are a compact and versatile option for music playback in a living room. They are typically designed to be used on a bookshelf or stand and can be paired with a small class-D amplifier to keep power consumption low.
Sealed enclosure design: A sealed enclosure design is a simple and reliable option for a bookshelf loudspeaker. It provides good low-frequency extension and can be designed to have a flat frequency response across the desired frequency range.
Two-way design: A two-way design with a woofer and tweeter is a common choice for bookshelf loudspeakers. It can provide good midrange and high-frequency performance, and the woofer can be chosen to provide good low-frequency response within the constraints of a sealed enclosure.
Use well-regarded drivers: While you don't want to use audiophile nonsense, it's still important to choose high-quality drivers for your loudspeaker. Look for well-regarded drivers that have a reputation for good performance at reasonable prices.
Consider crossover design: The crossover network, which divides the signal between the woofer and tweeter, is an important part of loudspeaker design. A well-designed crossover can help ensure a flat frequency response and good performance across the desired frequency range.
Overall, a sealed two-way bookshelf loudspeaker design with well-regarded drivers and a carefully designed crossover should be able to meet your requirements for a living room music playback system. Be sure to prototype and test your design before committing to manufacturing to ensure that it performs as desired.
....
Last edited:
As far as I can tell, it's currently an information aggregator with a pleasant user interface that can often repackage the ideas of others. It writes interesting text at times that makes people think it's thinking deeply, but it's not. Engineering/technical pursuits beyond those that have already been precisely described by someone else require a level of precision, understanding, and extrapolation that seem missing from these kinds of tools at this time.
https://www.newyorker.com/tech/annals-of-technology/chatgpt-is-a-blurry-jpeg-of-the-web
"Large language models identify statistical regularities in text. Any analysis of the text of the Web will reveal that phrases like “supply is low” often appear in close proximity to phrases like “prices rise.” A chatbot that incorporates this correlation might, when asked a question about the effect of supply shortages, respond with an answer about prices increasing. If a large language model has compiled a vast number of correlations between economic terms—so many that it can offer plausible responses to a wide variety of questions—should we say that it actually understands economic theory?"
. . .
"Let’s go back to the example of arithmetic. If you ask GPT-3 (the large-language model that ChatGPT was built from) to add or subtract a pair of numbers, it almost always responds with the correct answer when the numbers have only two digits. But its accuracy worsens significantly with larger numbers, falling to ten per cent when the numbers have five digits. Most of the correct answers that GPT-3 gives are not found on the Web—there aren’t many Web pages that contain the text “245 + 821,” for example—so it’s not engaged in simple memorization. But, despite ingesting a vast amount of information, it hasn’t been able to derive the principles of arithmetic, either. A close examination of GPT-3’s incorrect answers suggests that it doesn’t carry the “1” when performing arithmetic. The Web certainly contains explanations of carrying the “1,” but GPT-3 isn’t able to incorporate those explanations. GPT-3’s statistical analysis of examples of arithmetic enables it to produce a superficial approximation of the real thing, but no more than that."
. . .
"Even if it is possible to restrict large-language models from engaging in fabrication, should we use them to generate Web content? This would make sense only if our goal is to repackage information that’s already available on the Web. Some companies exist to do just that—we usually call them content mills. Perhaps the blurriness of large-language models will be useful to them, as a way of avoiding copyright infringement. Generally speaking, though, I’d say that anything that’s good for content mills is not good for people searching for information. The rise of this type of repackaging is what makes it harder for us to find what we’re looking for online right now; the more that text generated by large-language models gets published on the Web, the more the Web becomes a blurrier version of itself."
https://www.newyorker.com/tech/annals-of-technology/chatgpt-is-a-blurry-jpeg-of-the-web
"Large language models identify statistical regularities in text. Any analysis of the text of the Web will reveal that phrases like “supply is low” often appear in close proximity to phrases like “prices rise.” A chatbot that incorporates this correlation might, when asked a question about the effect of supply shortages, respond with an answer about prices increasing. If a large language model has compiled a vast number of correlations between economic terms—so many that it can offer plausible responses to a wide variety of questions—should we say that it actually understands economic theory?"
. . .
"Let’s go back to the example of arithmetic. If you ask GPT-3 (the large-language model that ChatGPT was built from) to add or subtract a pair of numbers, it almost always responds with the correct answer when the numbers have only two digits. But its accuracy worsens significantly with larger numbers, falling to ten per cent when the numbers have five digits. Most of the correct answers that GPT-3 gives are not found on the Web—there aren’t many Web pages that contain the text “245 + 821,” for example—so it’s not engaged in simple memorization. But, despite ingesting a vast amount of information, it hasn’t been able to derive the principles of arithmetic, either. A close examination of GPT-3’s incorrect answers suggests that it doesn’t carry the “1” when performing arithmetic. The Web certainly contains explanations of carrying the “1,” but GPT-3 isn’t able to incorporate those explanations. GPT-3’s statistical analysis of examples of arithmetic enables it to produce a superficial approximation of the real thing, but no more than that."
. . .
"Even if it is possible to restrict large-language models from engaging in fabrication, should we use them to generate Web content? This would make sense only if our goal is to repackage information that’s already available on the Web. Some companies exist to do just that—we usually call them content mills. Perhaps the blurriness of large-language models will be useful to them, as a way of avoiding copyright infringement. Generally speaking, though, I’d say that anything that’s good for content mills is not good for people searching for information. The rise of this type of repackaging is what makes it harder for us to find what we’re looking for online right now; the more that text generated by large-language models gets published on the Web, the more the Web becomes a blurrier version of itself."
Last edited:
This GPT answer of the first post resembles a mix of Parts Express projects. Dayton drivers, enthusiasm, a big three way and still not confident about its bass capabilities.
No, there is no standard size for all speakers.
New speaker technologies will eventually emerge, but each could be seen as just another plateau.
The box dimensions are not right or wrong, a "design" would specify crossover components and wiring diagram, individual chamber dimensions (the sealed back tweeter does not require a separate chamber) Fb, and would include bracing and stuffing details.
onaudio - you seem to have been addicted to the bot. Its just a learning software... is nice and fun n all but its not a revelation - no new insight or science is being made within its walls.. chill!! 🙂
//
//
How different is it from “computer” in Star Trek?
It seems pretty close at times.
A natural language interface to the sum of human knowledge would be amazing in and of itself.
Now, if it remixes in incorrect ways knowledge it found on the web, it’s just as good as useless.
The big question I have is: Is it in “learning mode” since it has been released to the public?
If it is, how will it differentiate between accurate interactions and deceitful interactions? (If a large majority of users extol the greatness of politician X, will it in fact take it for truth that politician X is indeed the best thing since sliced bread?)
Fascinating times…and their gadgets.
It would be easy to entirely dismiss the thing, but we have to realize very complex and useful software started with simplistic arithmetic machines that filled entire rooms.
It seems pretty close at times.
A natural language interface to the sum of human knowledge would be amazing in and of itself.
Now, if it remixes in incorrect ways knowledge it found on the web, it’s just as good as useless.
The big question I have is: Is it in “learning mode” since it has been released to the public?
If it is, how will it differentiate between accurate interactions and deceitful interactions? (If a large majority of users extol the greatness of politician X, will it in fact take it for truth that politician X is indeed the best thing since sliced bread?)
Fascinating times…and their gadgets.
It would be easy to entirely dismiss the thing, but we have to realize very complex and useful software started with simplistic arithmetic machines that filled entire rooms.
At a basic interface level, I can appreciate the comparison to the Star Trek computers. But I don't remember theirs convincingly stating false things because those words seem like they should go together, whether about bears in space or not.
I'm not trying to be dismissive about any of it - just trying to develop a clearer picture of where things seem to be at this time, based on my limited understanding of the tools. As a first draft generator for typical business/"news"/whatever text, GPT and similar tools seem to have potential. In scenarios where there are multiple skilled editors anyway, you may be able to replace a low-level writer with an AI, especially if it's tweaked to generate a certain style or corporate boilerplate.
After the initial hoopla, I was hoping for more logic/predictive ability, but the more I looked into it, the less likely that seemed for the current generations that are free-to-access and in the news cycle right now. Predictive models exist for various applications, of course, but normal people can't easily access them.
While GPT-type AI's may be an interesting advancement in how people interact with computers, the tools' basic inability to determine fact from fiction is a dealbreaker if you're trying to make important decisions based on their output. If you need to have enough knowledge to determine the output's veracity before using it, how much help is the tool to the average person?
A few more relevant bits from articles:
https://www.ft.com/content/e34c24f6-1159-4b88-8d92-a4bda685a73c
"In fairness, the tech’s creator is upfront about its limitations. Sam Altman, OpenAI’s chief executive, warned in December that ChatGPT was “good enough at some things to create a misleading impression of greatness . . . we have lots of work to do on robustness and truthfulness.”
. . .
"Meta even launched a science-specific text generator called Galactica. It was withdrawn three days later. Among the howlers it produced was a fictitious history of bears travelling in space.
Professor Michael Black of the Max Planck Institute for Intelligent Systems in Tübingen tweeted at the time that he was “troubled” by Galactica’s answers to multiple inquiries about his own research field, including attributing bogus papers to real researchers. “In all cases, [Galactica] was wrong or biased but sounded right and authoritative. I think it’s dangerous.”
The peril comes from plausible text slipping into real scientific submissions, peppering the literature with fake citations and forever distorting the canon. The journal Science now bans generated text outright; Nature permits its use if declared but forbids crediting it as co-author."
https://www.zdnet.com/article/chatg...y-job-but-theres-one-reason-im-not-panicking/
"To OpenAI's defense, the company states on the website that ChatGPT "may occasionally generate incorrect information" and has "limited knowledge of world and events after 2021." This makes sense because the AI doesn't always have a source of truth to reference; It doesn't know if the data it's pulling is true or false, especially in more complex situations"
I'm not trying to be dismissive about any of it - just trying to develop a clearer picture of where things seem to be at this time, based on my limited understanding of the tools. As a first draft generator for typical business/"news"/whatever text, GPT and similar tools seem to have potential. In scenarios where there are multiple skilled editors anyway, you may be able to replace a low-level writer with an AI, especially if it's tweaked to generate a certain style or corporate boilerplate.
After the initial hoopla, I was hoping for more logic/predictive ability, but the more I looked into it, the less likely that seemed for the current generations that are free-to-access and in the news cycle right now. Predictive models exist for various applications, of course, but normal people can't easily access them.
While GPT-type AI's may be an interesting advancement in how people interact with computers, the tools' basic inability to determine fact from fiction is a dealbreaker if you're trying to make important decisions based on their output. If you need to have enough knowledge to determine the output's veracity before using it, how much help is the tool to the average person?
A few more relevant bits from articles:
https://www.ft.com/content/e34c24f6-1159-4b88-8d92-a4bda685a73c
"In fairness, the tech’s creator is upfront about its limitations. Sam Altman, OpenAI’s chief executive, warned in December that ChatGPT was “good enough at some things to create a misleading impression of greatness . . . we have lots of work to do on robustness and truthfulness.”
. . .
"Meta even launched a science-specific text generator called Galactica. It was withdrawn three days later. Among the howlers it produced was a fictitious history of bears travelling in space.
Professor Michael Black of the Max Planck Institute for Intelligent Systems in Tübingen tweeted at the time that he was “troubled” by Galactica’s answers to multiple inquiries about his own research field, including attributing bogus papers to real researchers. “In all cases, [Galactica] was wrong or biased but sounded right and authoritative. I think it’s dangerous.”
The peril comes from plausible text slipping into real scientific submissions, peppering the literature with fake citations and forever distorting the canon. The journal Science now bans generated text outright; Nature permits its use if declared but forbids crediting it as co-author."
https://www.zdnet.com/article/chatg...y-job-but-theres-one-reason-im-not-panicking/
"To OpenAI's defense, the company states on the website that ChatGPT "may occasionally generate incorrect information" and has "limited knowledge of world and events after 2021." This makes sense because the AI doesn't always have a source of truth to reference; It doesn't know if the data it's pulling is true or false, especially in more complex situations"
Last edited:
If the Star Trek computer had been lying, no one could have told about it…because it is the very nature of these tools to dispense far more knowledge than anyone could hold. And, typically, people asking the questions don’t know the answers.
But experts in each field individually would have a field day proving it’s an automaton with little ability to discriminate.
But from a UX standpoint, I wish Google could be queried in this manner.
I do not think it’s revolutionary. But it does tickle my imagination.
Ironically, even humans cannot do that today on very basic topics.
In the populace and in the professional domain, where you will find incompetent people in pretty much every job. The incompetence of a surgeon doesn’t imply surgery should be banned entirely, though.
There is no denying the tremendous amount of bad advice found on the Internet today. This tool is a mere reflection of that and can be nothing else.
As for references, I will skip that in my reply.
We live in a world where no matter what you believe, you can find a “highly credible” pundit who corroborated your beliefs.
TLDR; it’s not heaven, it’s not hell, it’s merely a mirror of information found of the web.
While this is all mildly interesting, or infuriating depending on your own feelings about AI, I would caution anyone about trying to obtain too much technical info from ChatGPT. I recently saw a video on YT in which ChatGPT was asked to write a paper about a fictitious character from history and the thing completely made up the whole text AND references, all of which DID NOT EXIST! So when pressed on a topic ChatGPT seems to fake it. That is probably not how you want to design anything that you wish to actually work. For example:
https://news.ycombinator.com/item?id=33841672
https://news.ycombinator.com/item?id=33841672
ChatGPT is a conversational artificial intelligence system that leverages the power of generative models. At present, it functions by recognizing patterns in users' input and generating responses based on those patterns. Although this approach is efficient, it can result in false or nonsensical answers.
To address these limitations, we can use knowledge graphs and Long Short-Term Memory (LSTM) networks to enhance ChatGPT's performance. A knowledge graph is a graph-based data structure that represents concepts and their relationships in a given domain. LSTM networks are a type of recurrent neural network (RNN) that can store information over a prolonged period.
By combining ChatGPT with knowledge graphs and LSTM networks, we can improve its ability to understand the context of users' input, generate more accurate responses, and reduce computation time. ChatGPT can use knowledge graphs to identify the relationships between different concepts and leverage LSTM networks to remember previous interactions with users.
However, it is important to note that even with these improvements, ChatGPT can still provide erroneous responses. It is crucial to validate the information it provides and ensure that it aligns with domain-specific knowledge and common sense.
Last edited:
When AI spits out nonsense it is called "hallucination" (just learned this recently). It's a big problem for AI, actually. They are trying to come up with ways to encourage the learning algorithm to avoid this, but it might be an inevitable outcome of the current learning methods.
AI can produce pleasing results when used to make up fiction or art. But IMHO think the trend of using it to generate new factual information is fraught.
AI can produce pleasing results when used to make up fiction or art. But IMHO think the trend of using it to generate new factual information is fraught.
