audio bandwidth question

I’ve just started using Audacity, and I love it. I’m using El Capitan. I have a hopefully simple question though. I’m using it for audio recording, and it’s handy because I can export those recordings in many different formats (AIF, mp3, WAV, etc), AND at many diferent quality levels (kHz, kb/s). My question is about the quality level that Audacity uses to input that audio. Am I correct in assuming that “Project Rate (Hz)” at the far lower left of the home page is the sampling rate that is used for Audacity audio input? That is, if I want to export audio to mp3 with 128 kHz quality, I probably ought to set that Project Rate at least as high. Is that correct? I ask because that Project Rate seems to default to 44 kHz. More to the point, if I export at 128 kHz, and the Project Rate is at 44 kHz, the exported file quality will just be the lower of those two numbers?

Sample Rate and Compression Quality are very different.


The default project rate is 44100 Hz. That is , 44100 samples per second for each channel.
The default sample format is 32-bit float.
That means for a stereo track, the “bit rate” is 44100 x 2 (channels) x 32 (bits per sample) = 2822400 bits per second = 2756.25 kbps.

Exporting as 128 kbps second MP3 reduces the number of bits per second from 2756.25k to 128k.
The default project format in audacity is very much higher than 128 kbps MP3.

Thanks, but the answers to my questions are not obvious from that writeup. If I want a 128 kbs compressed audio file, what sample rate do I need to record it at to get full fidelity? What demands does compression quality make on sample rate? And also, is Audacity “Project Rate” in fact the sample rate?

I’m suspecting that if I want full AIF export quality, I’d better sample pretty fast to do justice to it.

I’d recommend working in 32-bit float “sample format” with a sample rate of 44100 Hz.

Ah, right. I was ignoring bits/sample! That makes perfect sense. So yes, the default “Project Rate” (which again, I am assuming, because no one has told me otherwise) is the sample rate, of 44 kHz is way more than 128 kbs.

Thank you!

There’s a bit about the project rate in the manual:

what sample rate do I need to record it at to get full fidelity?

MP3 and the other compressed formats (save one) work by throwing away some sound quality and you can’t stop it. It’s not as simple as making the sound muffled or unable to reproduce the sparkle of an instrument. There is no one-to-one relationship to compare the two values.

One of the clever tricks of MP3 (originally part of a video format) is to deliver a perfect sounding show by throwing away unimportant or inaudible sounds.

My silly joke is listening to two violins. One famous, old, terrific fiddle and one Tesco/7-Eleven. Nobody has any trouble telling the difference between them in real life, but if you run them through stiff MP3 conversion, you can’t tell which is which any more. MP3 assumed the quality overtones in the old fiddle were unimportant and threw them away.

99% of people listening to music like that don’t care.

And that’s why MP3 is so popular with everyone but published performers.

That compression dance has nothing or very little to do with sample rate. That does sound muffled if you get it too low. 44100 is the sample rate used in Music CDs. It’s very common which is one of the reasons it’s used in Audacity.


kbps is “streaming” audio. Nothing to do with audio files.

Of course, you can store streaming audio in a file too. How else would the server that “streams” the file over the internet get that to your computer.

Streaming is a necessary evil to make internet audio work. The server can switch on the fly to a lower bitrate if the connection deteriorates. Sound quality will be lower, but the music won’t cut out. That’s why “recording” from streaming sources is always less reliable than downloading.

MP3 is a streaming format that was especially developed for streaming. A nice side effect is that song files get smaller. That was very interesting when portable players only had very limited storage. MP3 isn’t “bad” an sich, but requires some work and a lot of experimenting to get right.

You can get some music to sound acceptable even at 64 kbps. Most music will require 128 kbps. And some music (jazz, fi) needs at least 192 kbps. At 320 kbps, you might as well use uncompressed wav format, as the reduction in file size isn’t as big anymore and you still lose some quality on “difficult” music.

When it comes to voice, it’s an entirely different picture…

When working with audio, streaming isn’t necessary. So you work with uncompressed formats, always. And these are simple:

  • When recording from a decent audio interface, use 24 bits, 44.1 KHz or 48 KHz. Forget about 96 or 192 KHz, unless you’re recording bats or something like that.
  • When playing back, 16 bits is enough. Once your editing/effects are done, export your work as 16 bits, 44,1 KHz (CD format), or 16 bits, 48 KHz (audio for use with video, DVD format).

Internally, Audacity works with 32 bits. Some DAW’s even work with 40 or 64 bits, to create extra room for calculations. But you won’t ever “hear” these extra bits, as there is no hardware to play back these resolutions. Just like Photosohop can use bitdepths no monitor can show…

You should know better than that. :wink: CD quality (16-bit, stereo, 44100 Hz) is 1411 kbps (16 * 2 channels *44.1 = 1411.2).


Yes, Gale, you’re right. Even uncompressed audio has a bitrate.

But hardly anyone uses that in pro audio, because it’s confusing. OTOH, iTunes and a lot of other players use it, as a way to disguise the difference between compressed audio and uncompressed audio.