WAV vs MP3: which is better, and when to use each

Neither is better for everything. A standard WAV file stores sound as uncompressed samples, so it is exact and large; an MP3 compresses sound by discarding detail, so it is small and lossy. Keep WAV while you record, edit, or join audio, and export MP3 last, when you share or deliver the file. Converting an MP3 to WAV makes it bigger but can't restore what the MP3 threw away.

The file sizes below are arithmetic on sample rate, bit depth, and bit rate. The Sume details come from the Timeline audio and Audio detach docs and the TTS 1.0 schema in the Sume API reference, all read on 2026-09-28.

Which has higher quality, WAV or MP3?

WAV, in the sense that it holds the audio exactly while an MP3 holds an approximation of it. An MP3 encoder decides which detail matters least and drops it to shrink the file; the higher the bit rate, the less it drops. However good an MP3 sounds, it holds less than the WAV it came from, and each time you edit an MP3 and save it as MP3 again, the new encode is another lossy pass.

Lossless compressed formats such as FLAC sit in between: smaller than WAV, with nothing discarded.

How much bigger is a WAV than an MP3?

A WAV's size is sample rate × bytes per sample × channels × seconds. An MP3's is its bit rate ÷ 8 × seconds. At 44,100 Hz and 16 bits, a stereo WAV is about 11 times the size of a 128 kbps MP3 of the same length.

Does converting MP3 to WAV improve quality?

No. Converting decodes the MP3 and writes its samples out uncompressed: the file grows to WAV size, but the detail the encoder discarded is gone for good. Convert only when a tool needs WAV. If you still have the original, export or generate the WAV from it instead of from the MP3.

Which should I use for AI voiceovers and music?

The same rule: keep WAV while the audio will be cut, joined, or synced to a face, and deliver MP3. Sume's Timeline audio docs describe MP3 output as smaller but say it re-adds priming padding at every edge, a short stretch of padding at the start and end of each file, so they tell you to keep WAV when the file will be joined again or drives lip sync.

Generated music that arrives as MP3 stays at MP3 quality: as above, converting it to WAV only makes it bigger, so convert only when a tool asks for WAV.

• Text to speech (TTS 1.0) returns MP3 at 44,100 Hz and 128 kbps unless you set output_format. For a WAV master, send "container": "wav" with "encoding": "pcm_s16le" and "sample_rate": 44100, the settings the API reference names for audio that will drive an avatar.
• A separate audio clip per sentence comes only with WAV or raw: for MP3, the sentence timings come back without the clips.
• Audio detach and Timeline audio already return WAV by default; Sume API output file formats lists every endpoint's options.