Why WAV files are so big
WAV stores uncompressed PCM audio: every sample is written out in full, with nothing removed. That makes it the format recorders and editors prefer, and it also makes the files large. A standard 44.1 kHz, 16-bit stereo WAV uses about 10 MB per minute. At 48 kHz, common on video and field recorders, it is closer to 11.5 MB per minute. A one-hour interview recorded in stereo is therefore around 600 MB, and a 24-bit or 96 kHz file is larger still.
MP3 at 192 kbps uses about 1.4 MB per minute, so the same hour comes out around 85 MB, roughly seven times smaller. At 128 kbps it is about 1 MB per minute and more than ten times smaller than the WAV.
Where these WAVs come from
Handheld and field recorders usually write WAV, typically 44.1 or 48 kHz at 16 bit. Interviews, lectures, rehearsals and ambient recordings all come off the memory card this way.
Audacity, GarageBand, Logic, Reaper, Pro Tools and other DAWs export WAV when you bounce a mix or a voiceover. Video editors export WAV when you render audio only. Some dictation devices and older phone systems also save WAV.
In every case the WAV is the right file to keep. It is the master. The MP3 is the copy you make when the file needs to go somewhere: an email attachment, a messaging app, a client review, a podcast host, a website, a USB stick for the car, or a form that will not accept a 600 MB upload.
Choosing a bitrate for WAV to MP3
Because WAV is uncompressed, this is the first lossy step for the audio. That is the best case for MP3: the encoder works from full detail, and at a sensible bitrate the result is very close to the original.
- 128 kbps: voice recordings, interviews, lectures, dictation and anything that will be transcribed. The smallest file, about 1 MB per minute.
- 192 kbps, the default: a good general setting. Music is fine for most listeners and speech has plenty of headroom. About 1.4 MB per minute.
- 320 kbps: music mixes, demos for clients and recordings you care about. About 2.4 MB per minute, still about a quarter of the WAV.
Mono, stereo and what the bitrate buys
At a constant bitrate the MP3 size depends only on length and bitrate, so a mono WAV and a stereo WAV of the same length give MP3s of the same size. The mono file simply has more bits per channel to spend, which is one more reason 128 kbps is comfortable for a mono voice recording.
Converting does not change the recording itself. Hiss from a cheap preamp, handling noise and a speaker too far from the microphone come through unchanged. If the WAV needs cleaning up, do that in the editor first and convert the result.
Large files, converted locally
The conversion happens inside your browser using FFmpeg compiled to WebAssembly. The WAV is read from your disk and encoded on your own machine. It is never uploaded, which matters twice over for WAV: a 600 MB file would take a long time to send to a server, and the recording stays private. RocketScribe receives nothing and stores nothing.
The first use downloads the engine, about 32 MB, which the browser caches; after that the page works offline. One file converts at a time. The practical limit is about 2 GB per file, enough for a few hours of stereo WAV on a desktop browser. Phones and tablets have less memory available to a web page and will fail on very large WAVs, so convert long recordings on a computer. The MP3 keeps the original name with an .mp3 extension and lives only in the open tab until you download it.
If you actually need the words
If the WAV is an interview or a lecture and what you want is a transcript, press Transcribe this file after converting. The MP3 goes into the transcription box below. That part runs on RocketScribe's servers, so it is uploaded, and the smaller MP3 uploads far faster than the WAV would. You get a timestamped, editable transcript in 90+ languages with exports to TXT, DOCX, PDF, SRT, VTT and CSV, and you can try it without an account.