Process generated tones in 32-bit float, then choose the export bit depth for the destination. For a finished 16-bit file, leave high-quality dither enabled. This Mac walkthrough explains those choices using a synthetic sine tone. Audacity’s bit-depth guidance.

Version matters here. Audacity’s current Mac download is 4.0.1, while its separate reference manual identifies itself as the Audacity 3.7 manual. The detailed preferences and export steps below apply to 3.7, whose Mac installers remain available on the official older-versions page. They should not be assumed to describe the Audacity 4 interface. Current Mac download, Audacity 3.7 Reference Manual, Older versions.

The task is to generate a short tone, change its amplitude with fades, and export a useful WAV file while avoiding unnecessary reductions in precision. No microphone, imported recording, or additional audio application is required.

Why use 32-bit float when the final file will be 16-bit?

The internal sample format controls how Audacity stores the track’s sample values during editing. Export encoding controls how those values are represented in the finished file. These are separate decisions: choosing a 16-bit delivery file does not require editing in a 16-bit track.

Audacity recommends its default 32-bit float format for processing. Floating-point storage retains precision through intermediate calculations and can represent sample values above 0 dBFS. That headroom is useful during processing, although the finished output still needs an appropriate level. Sample Format — Bit Depth.

The distinction matters even for a mathematically generated sound. Changing amplitude or applying a fade calculates new sample values. Audacity documents that processing a 16-bit or 24-bit track involves floating-point calculations followed by conversion back to the track’s format. Keeping the track in 32-bit float avoids those repeated reductions. Dither.

In Audacity 3.7, open Audio Setup → Audio Settings and check Default Sample Format. Use 32-bit float before creating your working track. Separately check Project Sample Rate, which sets the rate for newly generated tracks. For this example, use 44,100 Hz throughout; a destination that requests another rate takes precedence. Audio Setup Toolbar.

For an existing track, inspect its dropdown menu and Format submenu. Changing the default is not a substitute for checking the track you are actually processing. The track menu can convert between 16-bit integer, 24-bit integer, and 32-bit floating-point storage. Increasing the format later should not be treated as recovery of precision already discarded. Audio Track Dropdown Menu.

Which dither strategy matches the next use of the tone?

Dither randomizes the rounding errors introduced when reducing bit depth. Without it, rounding can produce patterns related to the signal. It adds a small amount of noise to make that conversion less objectionable; it does not restore the discarded numerical precision. Dither.

Use the next operation as the decision point. The following is a practical decision aid derived from Audacity’s format and conversion documentation.

Next use Format choice Dither decision
Continue processing in Audacity Keep the track in 32-bit float No bit-depth reduction is needed within that working track
Transfer to an editor that accepts floating-point WAV Export 32-bit float WAV Avoid reducing to integer PCM merely for the transfer
Deliver a finished 16-bit WAV Export 16-bit PCM from the float project Leave high-quality dither enabled; Shaped is the documented default
Deliver a requested 24-bit WAV Export 24-bit PCM from the float project This is also a reduction from float; leave high-quality dither enabled unless a specific numerical task requires otherwise

Audacity supports these WAV representations, but its documentation cautions that 32-bit float WAV will not play in many players. Use it when the receiving application accepts it. For ordinary delivery, follow the destination’s specified encoding. Export Formats supported by Audacity.

Set the conversion controls that affect the exported file

On a Mac running Audacity 3.7, open Audacity → Preferences → Quality. Its Real-time settings apply to playback conversion; its High-quality settings apply when writing converted audio to disk, including rendering and export. Changing playback dither is therefore not the way to choose export dither. Quality Preferences.

Leave Real-time dither at None for this workflow. Under High-quality Conversion, keep Shaped for a finished integer export. Audacity describes this option as concentrating noise toward higher frequencies. Triangle distributes it less strongly toward those frequencies. Rectangle is a documented option when extensive absolute silence matters because it does not add noise to zero-valued samples. These are conditional choices, not a listening-test ranking. Quality Preferences.

Generate and process a synthetic tone

Start in a new project so the generator cannot replace an existing selection. Use the following illustrative settings in Generate → Tone: Waveform: Sine; Frequency: 1000 Hz; Amplitude: 0.5; Duration: five seconds. Audacity’s generator accepts amplitude values from zero to one, and its duration control determines the generated length. Tone.

A sine wave keeps this example focused on a single frequency. Audacity also offers square and sawtooth waveforms, plus a square option designed to avoid aliasing. Those alternatives introduce waveform considerations that are unnecessary for choosing sample format and dither. Tone.

After generation, verify that the track’s Format is 32-bit float. Select the first 0.1 seconds and apply Effect → Fading → Fade In. Then select the final second and apply Effect → Fading → Fade Out. These durations are example editing choices. The documented linear fades move between silence and the selection’s original amplitude. Audio Track Dropdown Menu, Fades.

Keep the working track in float while making further changes. Save the editable project before exporting your delivery copy, so another encoding can be produced from the same working material.

A calculated example explains why quiet tone tails matter

Consider a separate illustrative sine wave with peak amplitude 0.0001, equivalent to −80 dBFS. Using normalized signed-PCM spacing of approximately 1 / 2^(bits - 1), that peak spans about 3.28 quantization steps at 16-bit and 838.86 steps at 24-bit. These are calculated values, not measurements from Audacity or results of a listening test.

The example makes the tradeoff concrete: a quiet generated waveform has relatively few integer levels available at 16-bit. A fade moves through progressively smaller amplitudes, so the conversion decision remains relevant even when the original tone started at a comfortable level. Audacity’s documentation identifies extremely quiet sounds as a situation where simple rounding can become more apparent. Dither.

Export the finished tone without an extra conversion

For a finished 16-bit copy in Audacity 3.7, check the high-quality dither setting immediately before export. Then use the following sequence. Export Audio.

  1. Open File → Export Audio and choose the computer export path if prompted.
  2. Choose a filename and destination folder.
  3. Select WAV as the Format.
  4. Select Mono for this single-channel tone.
  5. Set Sample Rate to 44,100 Hz, matching this example’s project.
  6. Set Encoding to Signed 16-bit PCM.
  7. Choose Entire Project unless you intentionally want only the current selection, then export.

WAV’s channel count, sample rate, and encoding are separate export controls. Check all three: the .wav extension alone does not tell you the bit depth. There is no need to convert the working track to 16-bit first; choose the delivery encoding in the export dialog. WAV Export Options.

If you also need a 24-bit delivery copy or a floating-point transfer file, create it from the retained working project. Avoid using the already reduced 16-bit export as the starting point for another version.

What if the export differs from what you expected?

First check the exported range and participating tracks. Audacity’s export documentation says that muted or otherwise excluded tracks are omitted, and hidden portions of clips are not exported. An unexpectedly short or incomplete file warrants checking those settings before changing dither. Export Audio.

If the tone changes pitch or duration after a rate adjustment, check whether you changed the track’s Rate rather than resampling it. In Audacity 3.7, changing Rate changes how quickly existing samples play. Resample changes the sample rate while retaining duration and pitch. Export also provides its own rate-conversion path. Audio Track Dropdown Menu.

For the next generated sound, keep the same sequence: establish the working format, finish processing, then make the delivery conversion from that project. This leaves the creative edits available and makes each exported file’s purpose clear.

Sources