My microphone can record at a 44100Hz and a 48000Hz sampling rate. I am then going to encode that audio to a WAV file with Audacity. Should I do 32 bit float, Signed 32 bit PCM, etc? What is the difference between the first two? Should I really encode at 32 bits or is 16 bits enough for high quality audio. I can even go up to 64 bits but I don't know if my microphone (Audio Technica ATR2500) supports it. Ah, I just looked it up and my microphone has a bit depth of 16 bits. What if I want to add 32 bit audio? Should I just go down to 16 bits? I am then going to compress this using the FLAC codec.
 
One more question, is there any point using 32 bit depth or higher? Do HD videos use 32 bits of depth?

Can anyone answer this?

Quote:

kensclark15 said:

My microphone can record at a 44100Hz and a 48000Hz sampling rate. I am then going to encode that audio to a WAV file with Audacity. Should I do 32 bit float, Signed 32 bit PCM, etc? What is the difference between the first two? Should I really encode at 32 bits or is 16 bits enough for high quality audio. I can even go up to 64 bits but I don't know if my microphone (Audio Technica ATR2500) supports it. Ah, I just looked it up and my microphone has a bit depth of 16 bits. What if I want to add 32 bit audio? Should I just go down to 16 bits? I am then going to compress this using the FLAC codec.
 
One more question, is there any point using 32 bit depth or higher? Do HD videos use 32 bits of depth?

Click to expand...

 
My quick and dirty non-scientific answer: 
This seems to be all about file size versus audio quality. Do you need to preserve everything in the original recording? If so, then you should use the highest available sampling rate for the microphone, and store in an uncompressed or a lossless compression format. Examples of lossless uncompressed are PCM and WAV. An example lossless compressed is FLAC. If you need a smaller file size still, then perhaps you have to throw a part of the audio away - they say it doesn't matter, you must judge if you can notice any difference. One example of a file format that does that is MP3 - the lower the MP3 bitrate the more of the audio that gets thrown away. More on file formats here: http://en.wikipedia.org/wiki/Audio_file_format
 
To add a bit more color:
It is helpful to appreciate bit rates: bit rate = (sampling rate) × (bit depth) × (number of channels). 44100kHz is a sampling rate. This means that every second, the audio is sampled 44100 times. You then get to bit depth. CDs, for example, are recorded using 16-bit, which means that for each sample, a number out of a possible 65,536 is calculated for the sound. I'll assume the number of channels is 2, i.e. stereo. So, how do we put this together with the above ... the file format PCM or WAV (and note that PCMs are stored in WAV files, it just uses a different algorithm to do the conversion than does a WAV-WAV, so to speak) store the recording in uncompressed lossless format, but you can still choose permutations to give a higher or lower bit rate. This means you could choose a WAV file, but use a low sampling rate to achieve a sub-optimal (if quality is what you want) result. The key point, again, is that you need to choose the bit rate to match your quality needs.
 
My final answer:
If you want ultimate quality, I believe the question is more one of digital versus analog. More here: http://www.drtmastering.com/faq2.htm#adtape. Regardless of that page, as you will understand from the bit-rate piece above, the general principle will be to use the highest sampling rates, and store everything in uncompressed file formats.
 
Does HD use 32-bit?
I believe not. See the audio comparison table here for more: http://en.wikipedia.org/wiki/Blu-ray_Disc#audio.
 
Note:
We haven't discussed the quality of the microphone etc here, but definitely worth being aware that the audio chain is only as string as the weakest link. Certainly, when you get to very high bit-rates, if something like a microphone is not up to the same standard, then the final recording will be blighted with inaccuracies.

So if the microphone records with a bit depth of 16 bits at a sampling rate of 48000Hz, and uses 1 channel (mono sound) and I add audio with 32 bits of depth with a sampling rate of 44100Hz and 2 channels (stereo), then I should store that in a 16 bit, 44100Hz sampling rate, and stereo WAV file? Then compress that using FLAC? I'm guessing I should do that because if I stored it at 32 bits of depth then it would be upscaled (whatever it's called)? Or would storing it at 32 bits of depth be fine? Like if 16 bit depth audio was stored at 32 bits of depth then would that degrade the quality? Also, would storing the WAV file using a 48000Hz sampling rate make it unsupported on some devices or will they downsample automatically?
 
I probably rambled on above but what I was trying to get at, what bit depth and sampling rate should I choose?

You can't upscale from 16 to 32, or anything else. Once the track is recorded at 16-bit, each of the many millions of samples will be a number from 0 to 65,535. The 24-bit ranges from 0 to 16,777,215. You can't go between the two.
 
So, what that means, to me, is that you have to record at high sampling and bit rates from the outset. I'd say you can store in FLAC because it will preserve the lossless side of things, but it is compressed so you'll save some space.
 
You mentioned 44,100 and 16-bit. That's CD format, which uses PCM. If a CD has the quality you want, shoot for recording to that standard, but just store it as FLAC rather than WAV or PCM.

Quote:

gerry123456 said:

You can't upscale ... once the track is recorded at 16-bit, each of the many millions of samples will be a number from 0 to 65,535.

Click to expand...

 
It's slightly more subtle than this. A DAC will upconvert, but it's through a process called interpolation, essentially it fills in the gaps. The important point is that there are gaps, i.e. degradation from the original has occurred, so you can't get back to 100% of the original, but you can come close.
 
It's kinda off your main point, but it's interesting (at least to me ): http://www.msbtech.com/support/How_DACs_Work.php

 As the mic outputs 16 bits, you won’t profit by using a larger value.
As you want to mix with other audio, almost all audio is in 44.1 kHz.
Use this value for your recording.
 
If you have sources with a different sample rate, you can always convert e.g. using SOX.
 
The moment you start to mix digital, you do calculations, mix 3 tracks and you add 3 numbers and divide by 3
1/3=0
Yes 0 because we are talking 16 bit integer.
You will have a substantial round-of error
100/3=33
1000/3=333
Etc.
The more bits you use, the lower the round off error (quantization error)
That is exactly the reason why Audacity offers you the option to store the project in 32 bits.
You import your 16 bit sources in Audacity, save the project in 32 and do all the mixing.
The final result you convert back to 16 (or 24 if you prefer) as almost no audio device supports 32 bits.

Songs in studios are now being recorded at 24bit, 96000hz.