16-bit vs. 32-bit for even volume levels

Hi -

I’m using Windows 7 and Audacity version 2.0.3.

I’ve been using (or trying to use) Sonar to create backing tracks for myself and my wife. I guess it’s my age (I’m 60), but Sonar has way too many bells and whistles that I don’t need and that make it difficult for me to use. Also, I’ve noticed that there can be quite a difference in overall volume levels from song to song.

I’ve only recently discovered Audacity and am now using it instead. Based on what I’ve been reading in the Audacity Manual, would exporting as a 16-bit .wav file act as a sort of compressor (I had been exporting as 32-bit in Sonar) and, if so, is there a way to import multiple 32-bit songs all at once and export them all as 16-bit files (before converting them to mp3s)?

I apologize if these are dumb questions, but as I’ve said I’m an old fogey. Back when we performed full time I created backing tracks using a Roland MC300 (on floppy disks - anybody remember those?), and that was even before General MIDI and Standard MIDI Files existed! Thanks.

No it wouldn’t.
“16 bit” refers to the file format - it means that there are 16 “bits” (1’s and 0’s) for each sample (point) along the waveform.
Audio CDs use 16 bits per sample and it is considered to be a high quality audio format.
“32 bit” is a lot more accurate than 16 bit as there are twice as many “bits” per sample. The extra precision is great while editing and processing the audio but it is overkill for the final product. Also, 32 bit float is not widely supported by other software, whereas 16 bit is supported by just about everything. It is therefore recommended that you use “32 bit float” while working in Audacity, and export the finished piece as 16 bit.

That is normal.
If you wish to make a compilation with all songs with a similar volume, the best way is to import all of the songs into one Audacity project, then use the “Solo” track button to listen to one track at a time.

Note that there is a maximum level that can be played without distortion (clipping). Attempting to amplify a track beyond “0 dB” (full track height) will result in unpleasant distortion.

The approach that I would take to making a CD compilation:

  1. Import all of the tracks for the compilation.
  2. Edit menu > Preferences > Tracks > Solo Button: set to “Simple”. (this allows you to quickly listen to just one track).
  3. Select all of the tracks (Ctrl+A)
  4. Effect menu > Normalize. DC offset correction enabled, Normalize to -1 dB.
  5. Use the track Solo buttons to switch from one track to another while playing (Note: If you do not Solo one track, all of the tracks will play at the same time, which will be very loud and very distorted).
  6. Identify which track is the quietest. We have already normalized all of the tracks close to “as loud as possible without distortion” so this track cannot be amplified louder. What we need to do is to reduce the level of all other tracks so that they are about the same loudness as this (quietest) one.
  7. Use the track Gain sliders to reduce the loudness of each of the other tracks so that they are about the same loudness as the quietest track.
  8. Use Export Multiple to export each of the tracks as a new file.

Thank you so much, Gale. Not only did you actually answer exactly what I was asking (something apparently unheard of on the Sonar forums), but you did so clearly and concisely so that I totally understand. Once again, thanks very much.

Actually you can give Steve credit for that answer. :wink:

Though if I was doing this a lot, personally I would make the gain adjustment automatically in other software using the Replay Gain algorithm.


Actually, 32-bit floating-point can go over 0dB. There is virtually no upper or lower limit with floating-point, which Audacity uses internally. If you convert a floating-point file with peaks above 0dB to 16-bit WAV, you’ll get clipping. Clipping is a kind of compression… A very BAD kind of compression. :wink:

And a quick note there is 32-bit and 32-bit floating. You want to be clear which one you’re using. Audacity uses 32-floating internally to avoid sound damage when you apply special effects and filters.