WAV, MP3, AAC, FLAC and Opus: Audio Formats Explained

Containers versus codecs, which format to convert to, bitrates for music and speech, iPhone ringtones, pulling audio out of video, and why MP3 to WAV adds no quality.

By · · Updated · 6 min read

Illustration for the Audio Format Converter guide

A 3-minute song as a 128 kbps MP3 is about 2.9 MB. Convert it to WAV and it becomes 31,752,044 bytes, roughly eleven times bigger, and it sounds exactly the same. That one fact explains most of the confusion around audio formats: some formats save space by throwing sound away, some save space without losing anything, and some do not try to save space at all.

This guide sorts out the formats you meet in practice, which one to convert to, how to choose a bitrate and a sample rate, and how to predict file sizes, with worked numbers from the Audio Format Converter and the MP4 to MP3 Converter.

Containers and codecs are different things

A file extension usually names a container, the wrapper that holds audio plus metadata. The codec is how the samples inside are encoded. The two are often confused because some names are used for both.

ExtensionContainerUsual codecLossy?
.wavRIFF/WAVEPCM (uncompressed)No
.mp3MPEG audio streamMPEG-1 Audio Layer IIIYes
.m4a, .mp4MPEG-4 (ISO BMFF)AAC, sometimes ALACAAC yes, ALAC no
.flacFLACFLACNo
.ogg, .ogaOggVorbis or OpusYes
.opusOggOpusYes
.webmWebM (Matroska)Opus, sometimes VorbisYes

So "convert M4A to MP3" really means "decode AAC, then encode MP3", and a .webm file from a browser recorder is almost always Opus audio in a WebM wrapper. When a player refuses a file, the codec inside is usually the problem, not the extension.

The formats in one paragraph each

WAV. Microsoft and IBM's RIFF-based format from 1991. The common form is a 44-byte header followed by raw PCM samples. Every editor, sampler, game engine and speech toolkit reads it, which is why it is the format of choice for editing and machine processing.

MP3. Standardized as MPEG-1 Audio Layer III in ISO/IEC 11172-3 (1993). It uses a psychoacoustic model to discard detail you are unlikely to hear. Its patents have expired, so MP3 is free to implement, and it plays everywhere. At 128 kbps it is acceptable; 192 to 256 kbps is transparent to most listeners.

AAC. The successor to MP3 in MPEG-2 and MPEG-4. At the same bitrate it generally sounds better than MP3. It is the default for Apple devices, YouTube and most streaming, usually in an .m4a or .mp4 container.

FLAC. Free Lossless Audio Codec, first released in 2001 and published as RFC 9639 in 2024. It compresses PCM to roughly 50 to 70 percent of its size and decodes back bit for bit. Use it to archive music without losing anything.

Opus. The IETF codec from RFC 6716 (2012). It covers everything from 6 kbps speech to high-quality music, has very low latency, and is what WebRTC calls, Discord and browser recorders use. In files it lives in Ogg (.opus, .ogg) or WebM. It only runs at 48, 24, 16, 12 or 8 kHz, so a 44.1 kHz source is resampled to 48 kHz on the way in.

Vorbis. Xiph.org's older free codec, usually in an .ogg file. Quality per bit sits between MP3 and Opus. Games and Linux software still use it widely.

AIFF. Apple's 1988 equivalent of WAV: the same uncompressed PCM, stored big-endian in an IFF-style container. Logic, GarageBand and many samplers default to it on the Mac.

WMA. Windows Media Audio, in Microsoft's ASF container. WMA 2 was the default for Windows Media Player rips in the 2000s. Today it matters only for old Windows software and devices, and no browser plays it.

M4R. Not a codec at all: it is an ordinary AAC .m4a file with a different extension, which iTunes, Finder and iOS treat as a ringtone. iOS limits ringtones to 40 seconds.

Which format to convert to

The right target depends on what happens to the file next. These are the nine outputs the Audio Format Converter offers, with the encoder behind each one:

OutputEncoderLossy?Good bitrateUse it for
MP3LAMEYes128 speech, 192 to 320 musicCar stereos, old players, anything that says MP3 only
M4AAAC-LCYes128 to 192Apple devices, phones, general listening
M4RAAC-LCYes192iPhone ringtones, cut to 40 s
FLACFLACNon/aArchiving music without loss
OGGVorbisYes128 to 192Games, Linux, free-software projects
OpuslibopusYes32 to 64 speech, 96 to 128 musicVoice notes, the web, small files
WAV16-bit PCMNo1,411 kbps at CD qualityEditors, speech models, game engines
AIFF16-bit PCMNoas WAVMac audio software and samplers
WMAWMA 2Yes128Legacy Windows Media setups only

Bitrate is the main dial for lossy formats. Speech is intelligible at much lower rates than music, especially in mono, so a podcast at 96 kbps mono MP3 often sounds identical to the same podcast at 192 kbps stereo. Opus stretches this further: 32 kbps Opus is fine for a voice memo.

WAV output at a fixed sample rate is written by the browser itself and needs no download. The other formats are encoded by ffmpeg compiled to WebAssembly, which the page fetches once (about 32 MB) from a CDN; the audio still never leaves your device.

Why WAV is so big: the arithmetic

PCM size has no mystery. Every second stores sample rate x channels samples of bit depth / 8 bytes:

bytes = 44 + seconds x sample_rate x channels x (bits / 8)

CD quality:   44,100 x 2 x 2 = 176,400 bytes/s  = 1,411.2 kbps
3:00 song:    180 x 176,400 + 44 = 31,752,044 bytes  (30.28 MB)
Speech model: 16,000 x 1 x 2 = 32,000 bytes/s
1:00 memo:    60 x 32,000 + 44 = 1,920,044 bytes     (1.83 MB)
Bar chart comparing the size of three minutes of audio: 2.9 MB as 128 kbps MP3, about 20 MB as FLAC, 31.8 MB as 44.1 kHz stereo WAV, and 5.8 MB as 16 kHz mono WAV
Bar chart comparing the size of three minutes of audio: 2.9 MB as 128 kbps MP3, about 20 MB as FLAC, 31.8 MB as 44.1 kHz stereo WAV, and 5.8 MB as 16 kHz mono WAV

Lossy codecs are described by bitrate instead: a 128 kbps MP3 uses 16,000 bytes per second regardless of content. FLAC is variable; quiet acoustic music compresses better than dense rock.

Lossy to lossless does not restore quality

This is the most common misunderstanding. Decoding an MP3 to WAV produces a perfectly faithful copy of what the MP3 contains, and the MP3 no longer contains what the encoder discarded. The WAV sounds identical to the MP3; it is just bigger and easier for other software to read.

The opposite direction has a cost too. Every lossy encode throws more away, so MP3 to AAC to MP3 degrades each time. Keep a lossless master (WAV or FLAC) and encode lossy copies from it, never from each other.

Good reasons to convert to WAV anyway:

  • An editor, DAW or game engine wants PCM.
  • A speech recognition model wants 16 kHz mono PCM.
  • You are about to edit and re-export, and want to avoid an extra generation of lossy encoding along the way.

Choosing a sample rate

The sample rate sets the highest frequency that can be stored: half the rate, the Nyquist frequency. 44.1 kHz captures up to 22.05 kHz, just above the roughly 20 kHz limit of human hearing.

RateTypical use
48 kHzVideo, film and broadcast; most phone and camera recordings
44.1 kHzMusic and CD audio
32 kHzSome broadcast and older digital recorders
22.05 kHzLow-bandwidth music previews, retro game audio
16 kHzSpeech recognition (Whisper, Vosk and most cloud APIs)

Upsampling a 16 kHz recording to 48 kHz adds bytes, not detail. Downsampling throws away high frequencies, which is harmless for speech and noticeable on cymbals.

Diagram of the conversion pipeline: the input file is decoded by the browser, resampled to the chosen rate, optionally mixed to mono, and written as 16-bit PCM with a 44-byte WAV header
Diagram of the conversion pipeline: the input file is decoded by the browser, resampled to the chosen rate, optionally mixed to mono, and written as 16-bit PCM with a 44-byte WAV header

Worked example: preparing a voice memo for transcription

You have a 1-minute voice memo from an iPhone as memo.m4a (AAC, 48 kHz, mono) and a transcription script that expects 16 kHz mono WAV.

  1. Open the Audio Format Converter, choose WAV, and set Sample Rate to 16 kHz and Channels to "Mix down to mono".
  2. Drop memo.m4a on the upload area. Safari, Chrome and Firefox all decode AAC.
  3. The result row shows something like M4A → WAV | 470 KB → 1.83 MB (the M4A size depends on its bitrate). The WAV size is exactly 60 x 16,000 x 2 + 44 bytes.
  4. Download memo.wav and pass it to your script.

The same steps in ffmpeg, if you prefer the command line:

ffmpeg -i memo.m4a -ar 16000 -ac 1 -c:a pcm_s16le memo.wav

For WAV the browser tool decodes with the browser's own Web Audio decoder and writes the file in JavaScript, so nothing is uploaded and nothing extra is downloaded. Pick MP3, M4A or Opus instead and the same file goes through the ffmpeg engine.

Getting the audio out of a video

A video file is a container with at least two streams, and the sound is usually AAC in MP4 and MOV files, Opus in WebM, and MP3 or AC-3 in older AVI files. Extracting it means copying or re-encoding that one stream and leaving the picture behind.

The MP4 to MP3 Converter does exactly that, and it is worth knowing which route it takes:

  • M4A from an MP4 or MOV. The AAC stream is copied as is into an audio-only MPEG-4 file. Nothing is decoded, so there is no quality loss, and an hour of video takes seconds. The ffmpeg equivalent is ffmpeg -i talk.mp4 -vn -c:a copy talk.m4a.
  • MP3 from anything. The AAC is decoded and encoded again with LAME. That is a second lossy generation, harmless at 192 kbps and above, but not free. The ffmpeg equivalent is ffmpeg -i talk.mp4 -vn -c:a libmp3lame -b:a 192k talk.mp3.
  • WAV from anything. Decoded once to PCM, ready for an editor or a transcription model.

A 52-minute lecture extracted as 96 kbps mono MP3 is 96,000 x 3,120 / 8 = 37,440,000 bytes, about 37 MB, which is small enough to keep on a phone. The same audio as 44.1 kHz stereo WAV would be about 550 MB.

Making an iPhone ringtone

An iPhone ringtone is AAC in a file ending in .m4r, no longer than 40 seconds. Choose M4R in the converter, type the point where the ringtone should start, and it cuts 40 seconds from there. To get it onto the phone, connect the iPhone to a computer and drag the .m4r onto the device in Finder on a Mac, or in iTunes or the Apple Devices app on Windows; it then appears under Settings, Sounds & Haptics, Ringtone. GarageBand on the iPhone can also export a ringtone without a computer.

Common mistakes

  • Renaming instead of converting. Changing song.wma to song.mp3 changes nothing inside. Some converters on the web have done exactly this and called it conversion; the file size staying the same is the giveaway.
  • Expecting WMA to work in a browser. No current browser decodes Windows Media Audio itself. The converter's ffmpeg engine can read it, so convert WMA to MP3 or M4A before sharing it.
  • Mixing up kbps and kHz. Bitrate (kbps) controls lossy quality and size. Sample rate (kHz) sets the frequency range of PCM. A "320 kbps WAV" setting means nothing.
  • Upsampling for quality. Going from 22.05 kHz to 48 kHz cannot invent the missing top end.
  • Forgetting mono. A stereo WAV of a single speaker is double the size for no benefit.

Pick the format by what happens next: FLAC or WAV for anything you will edit or archive, AAC or MP3 for listening, Opus for voice and real-time use, and 16 kHz mono WAV for machines that transcribe. For recordings made in the browser, the web speech guide explains why they arrive as WebM or M4A in the first place, and the Speech to Text tool is a quick way to make one.

Frequently asked questions

What is the difference between a container and a codec?

A file extension usually names a container, the wrapper that holds audio plus metadata, while the codec is how the samples inside are encoded. An .m4a file is an MPEG-4 container usually holding AAC, and a .webm file from a browser recorder is almost always Opus audio in a WebM wrapper. When a player refuses a file, the codec is usually the problem.

How big is a WAV file per minute?

A WAV file's size is 44 header bytes plus seconds x sample rate x channels x bytes per sample. At CD quality, 44.1 kHz stereo 16-bit, that is 176,400 bytes per second, so a 3-minute song is 31,752,044 bytes, about 30.28 MB. A 1-minute 16 kHz mono speech file is 1,920,044 bytes, about 1.83 MB.

Is FLAC better than WAV?

FLAC and WAV hold the same audio quality, because FLAC is lossless and decodes back to PCM bit for bit. The difference is size: FLAC compresses PCM to roughly 50 to 70 percent of its size, which makes it the choice for archiving music. WAV is uncompressed and read by every editor, game engine and speech toolkit.

Does converting MP3 to FLAC improve quality?

No. Converting a lossy MP3 to a lossless format such as FLAC or WAV produces a faithful copy of what the MP3 contains, and the MP3 no longer contains what its encoder discarded. The result sounds identical, only bigger. Keep a lossless master and encode lossy copies from it, never from each other.

What is the difference between kbps and kHz in audio?

Bitrate, in kbps, controls the quality and size of lossy formats such as MP3, AAC and Opus. Sample rate, in kHz, sets the frequency range of PCM audio: the highest storable frequency is half the rate, so 44.1 kHz captures up to 22.05 kHz. A "320 kbps WAV" setting therefore means nothing.

Sources and further reading

Open the Audio Format Converter

More Text/Speech Tools

Every tool in the Text/Speech Tools section, free and in your browser.