Audacity Forum

Please, can anyone give me any clues as to what I'm doing wrong?
I've mixed a number of podcasts in the past quite successfully. But something has gone wrong with the latest and I can't see how to retrieve the situation.
I'm using Audacity 1.3-13 beta on Windows XP to mix my own wav-format field recordings.
I've re-visited the mix because the sound quality of the mp3 export is so poor. I don't remember having any sound qualty problems when I fist did the mix (some months back). I originally exported as mp3 and then as wav. The sound quality on voices is pretty dire on both these export files and also when I play the mix in audacity. However, the sound quality is perfectly OK on the original wav files.
I've checked the mp3 export options and it was at 128kbps. So I upped this to 192 kbps and repeated the export of wav and mp3 versions.
There's no improvement. And it still sounds bad played in audacity.
On the web I found "How to Get Higher Audio Quality Using Audacity" and following this guidance, I've set the sample rate to maximum, Default Sample Format to 32-bit float, interface to -120dB. Still no improvement.
I've created a new project, imported one of the original wav files (without editing it in any way) and exported it as wav then mp3 - all without any loss of sound quality.
It seems there's nothing I can do to eliminate the distortion on the original mix. Is my only solution to start again from scratch, re-importing everything and re-mixing it all?

MP3 is a lossy compression format. In order to achieve a small file size, some of the data is thrown away and the remaining data is squashed together as tightly as possible. The discarded data cannot be recovered.

The greater the compression amount, the smaller the file, but the more data is discarded so the worse the sound quality.

Re-encoding an MP3 (that has already had some audio data thrown away) will throw away more data. Re-encoding always reduces the sound quality. You can't get back the data that has been discarded.

The settings in Audacity "Edit > Preferences > Quality" should be:
Default Sample Rate: 44100
Default Sample Format: 32 bit float

Unless you specifically need your MP3s to be "CBR" (Constant Bit Rate) then the "Presets" will generally give better quality. The "Standard" preset usually gives good results for music (or music and speech).

I fully understand that the original mp3 can't be retrieved. But why can't I re-create a new mp3 (with the correct settings)
by extracting from the original Audacity project? Has the aup file been irretrievably damaged by exporting the original mp3?

RWH wrote:why can't I re-create a new mp3 (with the correct settings)
by extracting from the original Audacity project?

You should be able to do that.
Does the Audacity Project sound good?

No. The sound quality is equally bad when I play the mix within Audacity.
It sounds the same regardless of what I do. But the original wav files (which I imported) sound perfect.
And my understanding is that these imported files are linked within the aup file
rather than copied into it.
I must have done something in Audacity to have corrupted the sound.
I've no idea what, so I've no idea how to undo it.

RWH wrote:And my understanding is that these imported files are linked within the aup file rather than copied into it.

They can be either. Uncompressed files (such as WAV files) can be either linked or copied (see here for a full explanation: http://manual.audacityteam.org/man/Audacity_Projects )
In what way does it sound bad?
Does the project have multiple tracks? If it does, try using the track "Solo" buttons so that you can listen to one track at a time - do the individual tracks sound bad?
Do other Audacity Projects sound OK?

The mix has lots of tracks. The distortion is most noticable on interviews, mostly on the male voices.
In some cases there's background noise superimposed (via another track).
But there's no improvement from playing the voice tracks solo.

I've realised one problem though. The original sound files are mp3, not wav - a mistake I made when doing the recordings.
I'd forgotten this.

All I can imagine is that I've caused the distortion by repeated application of the amplification effect, in trying to get
an even overall level.

Now the podcast is actually live on the net, I'm desperate to sort it out asap.
So I've started to replace the sound clips which are noticably bad by re-importing the original files.
It's messy because I've got to re-do the editing of these and I can never get them to quite the same length.
I guess the lesson is to check, check, check before going live, even when you've listened to is so much during the
editing that you're convinced it's all prefect.

Didn't reply to a couple of Steve's points.
The sound is bubbly - on interviews / sound-bytes but singing all seems OK.
There's no similar distortion on any of the other stuff I've done,
but all the previous ones were all recorded in wav. As mentioned, this was
recorded in mp3 by mistake.