So I was thinking about the current DAC/ADC system and how we are slowly moving towards lossless 24/96K + 24/192K recordings in the music industry. Before we just move on to the next best standard, shouldn't we consider making two steps forward and design a format that has as few limitations as possible? Some say that 24bit recordings are better then we could possibly need or want. While that was also said about vinyl, 8-track, tape, CD, color TV and HDTV. While the fact of matter is that progress is going to continue. ie. HDTV is already planned to be replaced by UHDTV by 2016-2020. 4k TVs are already in the consumer marketplace. Rather then drag our feet, why isn't the audio industry looking at designing an audio format that does away with as many limitations as possible? So how do we do this, IMO we jump straight to r 64bit floating point audio. We gain all of the benefits of floating point ie. accuracy, lower noise, higher dynamic range, greater resolution, the ability to record almost the highest level of dB possible on earth (192dB vs. 194db) at the cost of bigger files which will become a non-issue as data storage grows over the years. Aside from the fact that new hardware would need to be developed and that IC technology is not currently at the level of utilizing this whole range, yet, what would be the drawbacks of such a system? Once the media format stops being a limitation, what is the next limitation that would need to be worked on in the audio chain?
I predict that something like this will exist in the future. The final result will be losslessly compressed. My trademark for it is Audio Reality Recording . The file extension will be .arr so the pirates will adopt it quickly and make it the new standard. You heard it hear first folks.
I can't seem to find any data on a real world 64bit ADCs?. Is it possible to make one out of 2 32bit ADCs?
BTW please don't post in here just to say why this isn't worth the effort currently because of X. Imagine that we are 30 years in the future and X isn't a problem. A lot of things in audio/electronics/computers have improved greatly since 1982. Who is to say that the the problems of our current technology won't be mitigated in the near future. ie. 30 years ago TV was analog and had been color for only about 30 years, now we have 1080P and 4K.
Here is an interesting read related to the subject.
http://diwaves.com/tmp/sample_bits.htm
I predict that something like this will exist in the future. The final result will be losslessly compressed. My trademark for it is Audio Reality Recording . The file extension will be .arr so the pirates will adopt it quickly and make it the new standard. You heard it hear first folks.
I can't seem to find any data on a real world 64bit ADCs?. Is it possible to make one out of 2 32bit ADCs?
BTW please don't post in here just to say why this isn't worth the effort currently because of X. Imagine that we are 30 years in the future and X isn't a problem. A lot of things in audio/electronics/computers have improved greatly since 1982. Who is to say that the the problems of our current technology won't be mitigated in the near future. ie. 30 years ago TV was analog and had been color for only about 30 years, now we have 1080P and 4K.
Here is an interesting read related to the subject.
http://diwaves.com/tmp/sample_bits.htm
Last edited:
human hearing noise floor isn't likely to change short of tech redefining "human"
thermal/Johnson noise, Browinian motion of air molecules aren't going to yeild to any tech advance - there will always be limits on transducer and electronics S/N
any rationally gain structured 24 bit system will be naturally dithered by the random noise in the recording mics, much less the room/venue noise of any "real" musical performance
thermal/Johnson noise, Browinian motion of air molecules aren't going to yeild to any tech advance - there will always be limits on transducer and electronics S/N
any rationally gain structured 24 bit system will be naturally dithered by the random noise in the recording mics, much less the room/venue noise of any "real" musical performance
Last edited:
You are making a confusion between the packets of data that travels inside the PC and the size of the captured frames.
The processing path inside the studio consoles is as wide as 40 bit floating point to provide room for calculations.
But at analog interfaces, you don't have available more than 20-22 bit of real resolution. Noise gets in the way, human hearing gets in the way. Musical instruments have a limited dynamic range.
PS: Dither is needed only when you truncate the 24 bit 96kHz to 16 bit 44.1kHz to sell it on CD format.
The processing path inside the studio consoles is as wide as 40 bit floating point to provide room for calculations.
But at analog interfaces, you don't have available more than 20-22 bit of real resolution. Noise gets in the way, human hearing gets in the way. Musical instruments have a limited dynamic range.
PS: Dither is needed only when you truncate the 24 bit 96kHz to 16 bit 44.1kHz to sell it on CD format.
Last edited:
And yet we are progressing to 32bit.
HP-A8C | 32bit DAC Headphone Amplifier | Fostex
AK4399|Site of Asahi KASEI MICRODEVICES
SONIC_real_one - REAL resolution will improve with time, just as everything else does. Noise will diminish as we make newer and better designs. Musical instruments do have a limited dynamic range, electronic instruments theoretically don't. You sound like one of those people from the 60's who thought that vinyl, stereo and color TV was the end all be all, as we know that was a fallacy. Stop being stuck in the now and use a little bit of imagination.
Where will we be in 2042 audio-wise? At a minimum you can answer that linearly by saying, what has changed since 1982 and project that growth forward. The problem with that is that it doesn't take into account the move to the digital world which changes and grows at an exponential rate. Looking at this graph
Future | Computers | Internet | 2020 | 2030 | 2040 | 2050 | 2100 | 2150 | 2200 | 21st century | 22nd century | 23rd century | Far Future | Technology
I am trying to make an argument for us moving away from the universal shortsightedness that has crept into our culture because digital's rapid growth and to start thinking about developing standards that last the test of time. CD is dead, DVD-A and SACD are not going to supplant it, in fact all audio going forward will be digital transfered. MP3 is far from dead but the end is in sight, it has had a good 10 years but it will not have 10 more. FLAC or ALAC will be the next standard, or something that is similarly lossless. I just think that the next format needs to be designed for the future, not for planned obsolescence. As the internet has shown, planned obsolescence works great for the analog world, nothing really ever leaves the digital world.
HP-A8C | 32bit DAC Headphone Amplifier | Fostex
AK4399|Site of Asahi KASEI MICRODEVICES
SONIC_real_one - REAL resolution will improve with time, just as everything else does. Noise will diminish as we make newer and better designs. Musical instruments do have a limited dynamic range, electronic instruments theoretically don't. You sound like one of those people from the 60's who thought that vinyl, stereo and color TV was the end all be all, as we know that was a fallacy. Stop being stuck in the now and use a little bit of imagination.
Where will we be in 2042 audio-wise? At a minimum you can answer that linearly by saying, what has changed since 1982 and project that growth forward. The problem with that is that it doesn't take into account the move to the digital world which changes and grows at an exponential rate. Looking at this graph
An externally hosted image should be here but it was not working when we last tested it.
you can see we have increased the size of our hard drives by ~ 500,000 times in the past 30 years. In 30 years, as long as moore's law doesn't break down, a hard drive will be ~ 488 Petabytes.
Future | Computers | Internet | 2020 | 2030 | 2040 | 2050 | 2100 | 2150 | 2200 | 21st century | 22nd century | 23rd century | Far Future | Technology
I am trying to make an argument for us moving away from the universal shortsightedness that has crept into our culture because digital's rapid growth and to start thinking about developing standards that last the test of time. CD is dead, DVD-A and SACD are not going to supplant it, in fact all audio going forward will be digital transfered. MP3 is far from dead but the end is in sight, it has had a good 10 years but it will not have 10 more. FLAC or ALAC will be the next standard, or something that is similarly lossless. I just think that the next format needs to be designed for the future, not for planned obsolescence. As the internet has shown, planned obsolescence works great for the analog world, nothing really ever leaves the digital world.
So we have -174dBm@1hz to -134dBm@10khz for our resistors as our noise limits correct?
In 30 years we may have room temperature super conductors. Is this still a problem then?
Since glasses can cause violations in the Fluctuation-dissipation theorem - who is to say that we may not find an amorphous metal alloy that violates the theorem, which could be used to make ultra low noise electronics?
What if someone devises a way to compensate for johnson/thermal noise in a circuit in the next 30 years?
In 30 years we may have room temperature super conductors. Is this still a problem then?
Since glasses can cause violations in the Fluctuation-dissipation theorem - who is to say that we may not find an amorphous metal alloy that violates the theorem, which could be used to make ultra low noise electronics?
What if someone devises a way to compensate for johnson/thermal noise in a circuit in the next 30 years?
Really? You will improve the human hearing?
How about the termal noise of the air molecules? You will listen your recordings in a freezer? That limits the "silence" end.
Humar ear also has a threshold of pain (some 130dB SPL).
Both those limits ear performance for music at not more than 22-23 bit of dynamic - in a perfect silent room. Of course, you can consider a jet-engine at 1m as "music" and extend that range to some 24 bit, but that is it...
Last edited:
Room temperature superconductors might not help much, as you will probably still need resistors in your circuits. I suppose resistors might be simulated by using switched capacitors but then you have all the usual issues of sampling rates etc.
When people talk about the fluctuation-dissipation theorem and glassy systems they are not necessarily talking about transparent stuff from which windows can be made. I suppose it is conceivable that some substance might be found which never reaches thermal equilibrium with its surroundings and so has lower than usual thermal noise. Somewhat more likely is spurious claims from snake oil merchants.
Realistically, I think we have to accept that for domestic electronics we are probably stuck with Johnson noise for the forseeable future. And of course at audio frequencies there is flicker noise - still only poorly understood which could be either good or bad: good because a breakthrough could occur, bad because until we understand it we probably can't design it away.
When people talk about the fluctuation-dissipation theorem and glassy systems they are not necessarily talking about transparent stuff from which windows can be made. I suppose it is conceivable that some substance might be found which never reaches thermal equilibrium with its surroundings and so has lower than usual thermal noise. Somewhat more likely is spurious claims from snake oil merchants.
Realistically, I think we have to accept that for domestic electronics we are probably stuck with Johnson noise for the forseeable future. And of course at audio frequencies there is flicker noise - still only poorly understood which could be either good or bad: good because a breakthrough could occur, bad because until we understand it we probably can't design it away.
Just thought I would point out that the so-called 32-bit dacs linked above & mentioned from Asahi, ESS, etc., are NOT 32-bit dacs at all, but simply 24-bit dacs(arguable in itself in the AK4399's specs) with 32-bit digital filters included on-chip. All of thes 32-bit dac claims are ranging from misleading to downright fraudulent(e.g., Fostex's ad copy). Does an actual 32-bit dac for audio really exist? I haven't seen one.
To quote SY the second law is a bitch. A superconductor has 0 Ohms or it's not super anymore. In any case when the magic is found every device in the chain would have to use it.
Electronic instruments really don't have to be limited by thermal noise.
They can output digital data directly, with any practical bit depth, with valid signal right down to the lowest bit. They can synthesize their notes from coefficients rather than reading digitised samples from memory. The difference between calculating to 16 bit accuracy and calculating to 32 bit float accuracy is a simple matter of CPU horsepower.
Last edited:
Then again, for quite a long time amplifiers have been sold with distortion figures decades below what is needed to make the distortion inaudible, so why would marketeers care about the dynamic range of the human auditory system? If due to some technical breakthrough it would become possible to make 300 dB dynamic range audio equipment, there would probably be a market for it, albeit a small one.
Besides, it is not just marketing; as audio signals usually go through more than one piece of equipment, each apparatus has to introduce errors far below the threshold of audibility to prevent audible errors at the end of the chain.
Slightly off-topic: I think this is the biggest fault of subband coding formats like DCC, MP2, MP3, AAC et cetera. Lots of research were done to see what bit rate is needed to make the encoding artifacts inaudible for the average experienced listener after encoding and decoding one and only one time. But what happens when you listen to a news item of which the raw material is sent to the studio as an MP2 or MP3 file, edited, stored in MP2 format on the radio station's harddrive, mixed and then broadcasted as DAB?
Slightly off-topic: I think this is the biggest fault of subband coding formats like DCC, MP2, MP3, AAC et cetera. Lots of research were done to see what bit rate is needed to make the encoding artifacts inaudible for the average experienced listener after encoding and decoding one and only one time. But what happens when you listen to a news item of which the raw material is sent to the studio as an MP2 or MP3 file, edited, stored in MP2 format on the radio station's harddrive, mixed and then broadcasted as DAB?
You are confused, there is not editing of mp3/mp2 at broadcast/storing. Just in the final stage the audio file is compressed lossy.
Anyway, that is NOT what this discution here is about.
Short story long.... you can add countless bits after the 24-th one, they will be just noise and useless information because our hearing is what it is.
and no, 24 bit doesn't need any dither from the same reason - our hearing is what it is.
Anyway, that is NOT what this discution here is about.
Short story long.... you can add countless bits after the 24-th one, they will be just noise and useless information because our hearing is what it is.
and no, 24 bit doesn't need any dither from the same reason - our hearing is what it is.
Last edited:
Compressed content will continue to rule, just for the sake of convenience. Instead of holding higher fidelity versions of content, people will just hold more.
And handheld audio players today are barely hitting 16-bits, even in products with 24-bit codecs. "more quality" is not something you are hearing much about. In fact, things have gone backwards in some regards as people are using phones instead of dedicated players, and the phones are even worse. But everyone is happy carrying one less device
The state of the art codec you linked to is effectively 21 bits. Thus, in the last 25 years, we've added about 5 bits of dynamic range to state of the art. And 99% of the population cannot even hear that improvement. The next 25 years will likely not add another 5 effective bits.
Actually that "state of the art" DAC from above (AK4399) that claims 32 bit resolution, has -105dB THD+N. That is 18 bit real performance, worse performance than others, that, at 24 bit "only", can provide a real 20 bit performance (108-110dB THD+N).
PS: TI has also a 32 bit DAC that performs at the same level of -106dB THD+N (real 18 bit).
PS: TI has also a 32 bit DAC that performs at the same level of -106dB THD+N (real 18 bit).
Last edited:
I'm not confused at all. You are describing the "ideal" situation, but at broadcast stations, it is not unusual to store audio files in a lossy compressed format on systems like Dalet. It is also not unusual to use Musicam/MP2 compression to send audio over ISDN or IP links, from reporter to studio for example. And of course DAB broadcasting involves lossy compression, as does sending audio over a low-bit-rate studio to transmitter link to an FM transmitter. Anyway, it is indeed off topic.
I, for myself, like a reasonable amount of thermal noise from my tube amplifier. I also like the surface noise of the LP. And the candy paper noise in an auditorium. Music has nothing to do with them.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.