Reaching for Highest Quality possible in Audacity

So I am collecting data on how to strive for the Highest Quality possible in Audacity. So far, I have: setting it to 32-bit. I did a practice save, saving it to 32 bit Float, and when I looked at Properties, it wont tell me anything. Why is that? I am running Windows XP. Do I need to check my computer’s Sound Card? So many things. And whats the difference between 32-bit float and PCM? Does it matter? They appear in Options. What/where should I set my -db setting? Should you use a Dither? Can that dull frequencies? I dont want to do that. For high quality, should I Export using WAV file? I heard MP3 creates losses. Also, for going for high quality, is the mic ports on a laptop suficient for plug in, or should I use another port/connection on my com for quality, like a MiDi plug or port? Sorry to have so many ???'s but hope this thread will prove useful to those who are just getting the hang of things. And if anyone can ADD anything to ensure quality, please do.

Also, for going for high quality, is the mic ports on a laptop suficient for plug in, or should I use another port/connection on my com for quality, like a MiDi plug or port

MIDI is the connection your computer uses to push a key on your musical keyboard. That’s it. If your keyboard makes a sound that’s almost irrelevant. The computer pushed the key and now it can pack up and go home. It’s machine control, not a sound format.

Since you’re on Windows, the analog connections on your laptop are almost certainly trash. The Mic-In on many Windows Laptops is noisy, easily overloaded, distorted, and mono.

A good quality stereo alternative is the UCA-202 that we frequently recommend. We have a handful at work and I own one. It’s not top quality, but it gets really close.

All that detail in the post and we know zero about the show. The type of show and goals make a really big difference how many hoops you have to jump through to get there. Are you live capturing your grandmother telling stories from her Irish upbringing?

So I am collecting data on how to strive for the Highest Quality possible in Audacity.

What’s the budget? We can do amazing things in Audacity but if your outside world connection is terrible, it won’t benefit anybody at all and just wastes space.


What will be the intended audience of your finished product? Most work done in Audacity is intended to be listened to by humans. So the ultimate limiting factor on quality is the human hearing system. Undamaged human hearing is generally accepted as encompassing the frequency range 20Hz to 20Khz. There is, therefore, little point in capturing frequencies outside these limits.

The sample rate (for example: 44.100) is what defines the range of frequencies that can be captured. An accepted broad generalization is that, to capture a specific frequency you need to use a sample rate that is twice that frequency. On that basis, 44100 will capture the 20KHz that is the top end of perfect human hearing.

Considerations such as 16-bit, 24-bit or 32-bit sample format affect the accuracy with which the analogue sound file has been digitised. 16-bit and 24-bit systems store the data as integers (whole numbers), whereas 32-bit systems store the data in floating point arithmetic. Thus a 32-bit system can capture subtler nuances. But can the human ear detect those differences?

Putting both those paragraphs together explains why you will find so many references here on the forum to setting the Audacity project rate to 44100 and letting Audacity use its default sample format of 32-bit Float. It also explains why 44100 and 16-bit is the combination chosen for commercially produced CDs. It is a combination that can capture the full range of typical human hearing and deliver the majority of the nuances that those human ears can detect.

That’s the technical/technology side of quality. After that, quality becomes more subjective; i.e. does it sound good to you? The hardest part of audio then becomes getting a good clean original recording that has no unwanted sounds and is neither under-recorded nor over-recorded. Now we’re in the realms of choosing and using a microphone or other digital recording device. Speaking purely personally, I would never even dream of attempting to feed an analogue signal into a computer and get a good clean recording. The inside of a computer system is “electronically hostile” to very low level analogue signals. There are too many sources of unwanted electro-magnetic frequency emission and too many feet of unshielded cable capable of acting as an aerial to pick up that electronic noise. I always recommend keeping the analogue part of the recording process well away from the computer.

Finally we come to mixing: a black art if ever there was one. Now we are really in the realms of the subjective. I can vouch only for mixing the spoken word with music and natural sounds, and the only solid advice I can give is that to keep the spoken commentary clear above the rest of the soundtrack I need to have about -20dB of separation in the sound levels.

Hope this has been of some help to you.

The other thing you have to cosider is what you are going to play the resultant material on.

  1. I don’t know of players taht will pay back 32-bit WAV files

  2. If you are planning to burn CDs then you will need to export 44.1 kHz 16-bit PCM stereo WAV files (but you will probably still want to work in Audacity at 32-bit float if you are doing editing - and then you should be using some form of dither, shaped or triangular, on the downsample 32-16 bits). If all you are doing is simply recording an LP and exporting the tracks without any editing then you could consider working in 16-bit in Audacity and thus avoid the needfor downsampling on export - but most folk want to do some editing clean-up on their LP captures.

  3. If you are going to be playng back on a portable device like an iPod say then you can use WAV for highest quality playback (iPods will definitely play 16-bit WAVs but not 32-bit WAVs AFAIK) - but this will restrict the amount of music that you can get on the portable device (which kind of defeats the object). On my iPod I use compressed audio files - I export WAVs from Audacity and use iTunes to convert to AAC 256 VBR which sound pretty good to my ageing ears (on my studio qualy headphones not the silly little earbuds).

  4. If you are planning to burn DVSs you will need to work and export in 48 kHz

If you are creating a serious archive the recoding at 32-bit float and exporting those WAVs at 32-bit is probably the way to go, probably exporting the raw capture file. This should give you a raw file which can be reprocessed on future (possibly improved) audio editing applications. Personally I don’t bother - but I do retain backup copies of the production 16-bit WAVs that I make.



A good quality stereo alternative is the UCA-202 that we frequently recommend.

You may like to add the following link to your recommendation for the UCA-202:-

I have not read the complete review but I do agree with what he says about the blind testing.

Thanks for that reference.

Nobody claimed it was the best you could possibly do, but just the fact that many Windows laptops don’t do this at all put this device directly in the sweet spot of price/quality. It also appears in the Multitrack/Overdubbing tutorial as one of the devices which will manage live monitoring both directions.

The fact that it will not blow your Koss headphones across the room is largely irrelevant.

If you have a Mac and then most times you can leave your UCA-202 at home. You do still need one for overdubbing, though.


As I recall there as a big kerfuffle about NT6 (aka Vista) and the way it “broke” a lot of audio software, because the audio subsystems in Vista (which are the same in NT6.1, aka “Windows7”) are very different from the old ones in NT5 (Windows2000) and NT5.1 (XP).

XP’s “best quality sound” will be CD quality. A later OS will provide DVD-quality sound, and yes, there is a difference if your systems and ears are good enough to hear it.

Not to throw out babies in bathwater, but since the original question was how to get the BEST sound quality…First, one has to throw out XP and get a newer OS.

Which brigs up smartly back to:

What’s the budget?

When they say no limits, posters usually mean no limits as long as it’s free.


Actually, no. The original question was…

the Highest Quality possible in Audacity.

But nobody can resist blowing that out to the uselessness of making Audacity perfect and leaving the rest of the system a mess.

To specifically answer you, Audacity defaults internally to 32-bit floating format and unless you change it, that’s the very best you can do and it’s excellent. You might bump the rate up to 48000 instead of 44100 (DVD instead of CD) to get rid of that last little ambiguous sampling error near the top As for getting in and out of Audacity, that will kill you. See the rest of this thread.