How does digital audio even work? (Specs, protocols, error handling, clocks??)

Jump to Latest
#21
In the last graph created by the spreadsheet, I had increased the time (x) value by 0.5 each time. This time I have set the increment to 0.1, which means a higher sampling rate of 10 times a second as opposed to twice every second.

Not the smoother curve in the same time interval: 0 to 14 seconds. For the purposes of digital music, a sample rate of 44100 samples per second is sufficient in order to represent the highest frequencies.

image_2022-03-20_105529.png
 
#22
jan.didden said:
That's the transportation part of it, that comes later. He is nicely describing the generation/storage.

Jan
Click to expand...

Thanks. This series of posts is a result of my trying to learn about digital music, and for me the best way to learn is to proceed from the basics and slowly build my knowledge. It is posted here with the help it will be of use to others. The existing explanations are quite complex, maybe necessarily so, and this approach seems to be useful. I will post relevant links - other resources will be used.

Please feel free to correct me if I make any mistakes.
 
#25
The original question was about transmission of data, yes. Maybe I could have started a different thread. Since we are here, I post a few links:

Next, we look at bit depth. This is how many amplitude values are recorded and stored in the system:

There is one more important factor to consider when considering the sampling process: bit depth. Bit depth represents the precision with which the amplitude is measured. In the same way that there are a limited amount of samples per second in a conversion process, there are also a limited amount of amplitude values for a sample point, and the greater the number, the greater the accuracy. A common bit resolution found in most standard digital audio systems (Hi-Fi, Compact Disc) is 16 binary bits which allows for a range of 65536 (2^{16) individual amplitude values at a point in time. - Wikibook
Click to expand...
The question is what is the bit depth for mp3s? mp3 files are different, and they are created to contain a certain number of bits per second. An mp3 file, which has some information removed to save storage space and and transmission bandwidth, presumably, will have to have fewer discrete values for amplitude, however we need to refer to the way mp3 is encoded. There are lossless audio signal digitization methods also, it has to be mentioned.

As Wikipedia states:

When performing lossy audio encoding, such as creating an MP3 data stream, there is a trade-off between the amount of data generated and the sound quality of the results. The person generating an MP3 selects a bit rate, which specifies how many kilobits per second of audio is desired. The higher the bit rate, the larger the MP3 data stream will be, and, generally, the closer it will sound to the original recording.
Click to expand...

https://en.wikipedia.org/wiki/MP3

Mods: Do I need to start a new thread?
 
#26
Experimenting with sampling rates, I used Audacity to change the sampling rate to 8,000 from 44,100 and exported the song as an mp3. File size reduced from 5.2 mb to 1.0 mb, but sound quality was absolutely terrible, with no highs and distortion all around.

Similarly changing the bit rate for a .wav file of 43 mb reduced the size of the file to 8 mb, however, again, distortion was high and it was not pleasant.
 
#28
Mb Yes. Considering the title, I don't think I need to change threads.

I just watched this video on ADCs. There are many types, but ADCs apply to the recording studio side of the equation. If your PC has a microphone input, there will be an ADC involved as well.


Is it possible to write a software ADC by accessing the raw data from the microphone port?
 
#29
BasicHIFI1 said:
Experimenting with sampling rates, I used Audacity to change the sampling rate to 8,000 from 44,100 and exported the song as an mp3. File size reduced from 5.2 mb to 1.0 mb, but sound quality was absolutely terrible, with no highs and distortion all around.

Similarly changing the bit rate for a .wav file of 43 mb reduced the size of the file to 8 mb, however, again, distortion was high and it was not pleasant.
Click to expand...
Reduce the SR to 8kHz reduces your BW to less than 4kHz. And no subsequent MP3 encoding will transform that into HiFi. Even without any encoding, 8kHz sample rate sounds bad.
The encoding, MP3 or FLAC or whatever, is another, but different method to save the audio stream BW.
They should be considered separately and not mixed up.

Jan
 
#32
jan.didden said:
Reduce the SR to 8kHz reduces your BW to less than 4kHz. And no subsequent MP3 encoding will transform that into HiFi. Even without any encoding, 8kHz sample rate sounds bad.
The encoding, MP3 or FLAC or whatever, is another, but different method to save the audio stream BW.
They should be considered separately and not mixed up.
Click to expand...
I was experimenting with sample rate reduction and file sizes. Of course once the information was gone, it did not matter what format I exported into . So the reduction in sampling rate is a parameter to adjust for smaller file sizes.

Audacity has a constant kbps setting for exporting mp3 files, which translates into a fixed sampling depth, what is the bit depth of a 192 bps file then?

Decent sound quality even at constant 128 kbps, which means some of my mp3s were not good at all. 4.2 MB compared to the 44 MB original. Not bad.

1647769550518.png


For purposes of illustration.
 
#34
Thanks, reading it now. Yes DSP is on my wish list someday. Phase corrections and all.. never heard a DSP in action.

Skimmed through it, clear diagrams and explanation, however slide 53 onward is a bit difficult to follow, will have to re-read.

With the flexibility of software, I have been asking if it is possible to implement the 'processing' part of the DSP through software only, are there latency issues? Is DSP performed through hardware with software helping configure it? Is it possible to also implement the ADC and DAC functions through software only or maybe simulate it?
 
Last edited:
#35
BasicHIFI1 said:
Thanks, reading it now. Yes DSP is on my wish list someday. Phase corrections and all.. never heard a DSP in action.
Click to expand...
My main motivation for the miniDSP FLEX is that it comes with Dirac Live!. I've done a review for a NAD that came with DL and was impressed.
So I'd do a DL correction, then 2-way crossover and send the result to the ADI-2 Pro.
I have the ADI-2 Pro but the FLEX availability is sill not confirmed, Expected end of 1st quarter. Which is soon.

Jan
 
#37
BasicHIFI1 said:
...slide 53 onward is a bit difficult to follow...
Click to expand...

Slide 53 is a continuation of the topic in slides 51 and 52, in the section called 'Aliasing.' Its that one set of sample points can represent an infinite series of sine waves. 'Alias' means there is more than one name for the same thing, as when a person uses more than one name. In this case, many different frequencies can be 'the same name' for one set of sample points. Slides 51 and 52 illustrate how that can happen. As a result, an anti-alias filter is required before the ADC. That allows us to be sure of what the actual input frequency was.

At the output of the DAC a reconstruction filter is also needed. Another way of looking at the process of reconstruction is that its connecting the samples (dots) with a smooth curve with exactly the same smooth shape as the original analog signal before it was digitized. In oversampling DACs it turns out that most of the reconstruction filtering can be done digitally.
 
Last edited:
#38
BasicHIFI1 said:
I was experimenting with sample rate reduction and file sizes. Of course once the information was gone, it did not matter what format I exported into . So the reduction in sampling rate is a parameter to adjust for smaller file sizes.

Audacity has a constant kbps setting for exporting mp3 files, which translates into a fixed sampling depth, what is the bit depth of a 192 bps file then?
Click to expand...
An mp3 does not necessarily reduce bit depth or sampling rate. I'm guessing it uses 16 bits (the original signal for generating mp3s is usually the CD equivalent, 16 bits/44.1kHz), but I don't know. If you really want to hear what an mp3 does, don't reduce the sampling rate (that flat-out reduces bandwidth as mentioned), but make mp3s at 64 and 96, and hear how they sound next to 128, 192 and the original .wav file. mp3 is a LOSSY format, meaning it's never going to sound exactly like the original (though higher bitrates like 256 and above come awfully close, I certainly can't tell the difference). FLAC (and a few other formats) is lossless and when played back will sound exactly like the original CD or .wav file, though FLAC files are about half the size.

But yes, all these different things (CD, mp3, FLAC) are basically different technologies.
 
You must log in or register to reply here.
Top Bottom