Audio Formats and Conversion, Explained
After reading this you will know what actually happens when you convert a WAV to MP3, why a lossy re-encode never gets cleaner, and how to pick a bitrate that holds up without wasting space.
What audio conversion actually does
A digital audio file is a long list of numbers. Each number is a sample: the position of the speaker cone at one instant. A CD stores 44,100 samples per second per channel, each sample a 16-bit integer between -32768 and 32767. That raw stream is what a WAV file holds, plus a small header.
Converting between formats means one of two things. A container swap (WAV to FLAC) keeps every sample bit-for-bit and only changes how the numbers are packed. A codec change (WAV to MP3) discards some of the numbers on purpose, using a model of what your ear cannot hear, so the file gets smaller.
Concrete example: a 3-minute stereo track at 44,100 Hz and 16 bits is 44100 × 2 × 2 × 180 = 31,752,000 bytes, about 30.3 MB as WAV. Encode it to MP3 at 192 kbps and it becomes 192000 ÷ 8 × 180 = 4,320,000 bytes, about 4.12 MB. That is a 7.4:1 reduction, and most listeners cannot tell the difference on ordinary speakers.
Lossless versus lossy, and when each wins
The formats split into two groups.
- Lossless (WAV, FLAC)
- Every sample is preserved exactly. FLAC compresses by predicting each sample from the ones before it and storing only the small prediction error, so a quiet or repetitive track shrinks more. Typical FLAC output is 50% to 60% of the WAV size with zero quality change.
- Lossy (MP3, OGG Vorbis, AAC/M4A)
- The encoder throws away detail your ear is unlikely to notice, then stores what remains at a target bitrate. Smaller files, but the discarded detail is gone for good.
Use lossless when the file is a master you will edit or re-encode later, or when you are archiving. Use lossy when the file is a final delivery for listening and size matters, such as a podcast download or a music library on a phone.
Converting from one lossy format to another (MP3 to AAC, say) decodes the already-damaged audio and damages it again. Each generation loses a little more. Always keep a lossless master (WAV or FLAC) and encode fresh copies from it.
Bitrate: the one number that controls lossy size
For a constant-bitrate (CBR) lossy file, size is almost entirely fixed by bitrate and duration. The formula is direct.
Here S is the file size in bytes, b is the bitrate in bits per second, and t is the duration in seconds. Divide by 8 to turn bits into bytes. A 4-minute song at 128 kbps is 128000 × 240 ÷ 8 = 3,840,000 bytes, about 3.66 MB. The same song at 320 kbps is 9.16 MB.
Bitrate does not change with sample rate or channel count in CBR mode; those choices affect quality at a given bitrate, not the size. Halve the bitrate and you halve the file. Variable bitrate (VBR) spends more bits on complex passages and fewer on silence, so the average lands near your target but the exact size varies.
Choosing a bitrate that holds up
More bits mean the encoder discards less, but the payoff flattens. Below is a rough guide for stereo music at 44,100 Hz. "Transparent" means most listeners cannot distinguish it from the lossless source in a blind test.
| Bitrate | Size per minute | Quality |
|---|---|---|
| 96 kbps | 0.72 MB | Audible artifacts on music |
| 128 kbps | 0.96 MB | Acceptable, common for speech |
| 192 kbps | 1.44 MB | Good, hard to fault |
| 256 kbps | 1.92 MB | Near transparent |
| 320 kbps | 2.40 MB | Transparent for nearly everyone |
The size-per-minute column comes straight from the formula: 192 kbps gives 192000 × 60 ÷ 8 = 1,440,000 bytes = 1.44 MB. AAC (M4A) is more efficient than MP3 at the same bitrate, so AAC at 128 kbps sounds roughly like MP3 at 160 kbps.
A worked conversion using the demo defaults
WAV master to a 192 kbps MP3
Take a 3-minute (180 second) stereo WAV recorded at 44,100 Hz, 16-bit, and convert it to MP3 at the default 192 kbps.
- Source size:
44100 × 2 channels × 2 bytes × 180 s = 31,752,000bytes, about 30.3 MB. - Target MP3 size:
192000 × 180 ÷ 8 = 4,320,000bytes, about 4.12 MB. - Compression ratio:
31,752,000 ÷ 4,320,000 = 7.35, so the MP3 is about 13.6% of the WAV size. - Sample rate and bit depth stay at 44,100 Hz. The 16-bit integers are decoded to the encoder's internal representation, filtered, and repacked at the new bitrate.
- What you lose: frequency content the psychoacoustic model judges masked or inaudible, mostly high-frequency detail and very quiet sounds sitting next to loud ones.
Run the same file to FLAC instead and you would expect roughly 15 to 18 MB, lossless, with every sample intact.
Reading the result and checking it
After a conversion, sanity-check three things. First, the size: it should match S = b \cdot t / 8 within a few percent for CBR, or vary for VBR. If your 192 kbps 3-minute file came out at 12 MB, something is wrong (probably VBR at a much higher quality setting, or the wrong bitrate applied).
Second, the duration should be unchanged. Conversion never alters playback length; if it did, the sample rate was changed by mistake.
Third, listen on the gear you actually use. Artifacts that vanish on laptop speakers can be obvious on good headphones: a "swirling" or "underwater" texture on cymbals and sibilant "s" sounds is the classic low-bitrate MP3 tell.
The conversion engine (ffmpeg compiled to WebAssembly, about 25 MB) downloads once on first use, then stays cached in your browser. Everything runs locally, so your files are never uploaded.
Common mistakes
Re-encoding a lossy file "to save space." Taking a 320 kbps MP3 down to 128 kbps does shrink it, but you are compressing already-compressed audio and the artifacts stack. Encode from the lossless master if you still have it.
Upsampling to "improve" a file. Converting a 128 kbps MP3 to WAV or to 320 kbps MP3 makes a bigger file with exactly the quality of the 128 kbps source. The lost detail does not come back. You cannot add information that was thrown away.
Ignoring channel count. Converting stereo speech to a 192 kbps file wastes bits. A voice recording is fine at 96 kbps mono, which is a quarter the size of 192 kbps stereo.
Assuming all lossy formats are equal. At 128 kbps, AAC and OGG Vorbis generally sound better than MP3 because they are newer designs. If your target device supports M4A/AAC, prefer it over MP3 at the same bitrate.
Related tools
If your audio is trapped inside a video file, use Extract Audio from Video to pull the track to MP3, WAV or FLAC first, then convert it here if you need a different format. For the video side, Video to GIF and GIF to MP4 handle the moving-picture conversions.
Frequently asked questions
Does converting WAV to FLAC lose quality?
No. FLAC is lossless. Every sample is reconstructed bit-for-bit when you decode it. You only gain a smaller file, typically 50% to 60% of the WAV size.
What bitrate should I use for a podcast?
For a spoken voice, 96 kbps mono is usually plenty and keeps downloads small: about 0.72 MB per minute. If you have music beds, step up to 128 kbps.
Why does my MP3 sound worse after two conversions?
Each lossy encode discards detail. Convert MP3 to another lossy format and the encoder works from the already-damaged audio, then damages it again. Keep and encode from a lossless master to avoid this.
Can I make a low-bitrate file sound like the original by converting up?
No. Converting a 128 kbps MP3 to 320 kbps or to WAV only grows the file. The information removed at 128 kbps is gone permanently.
Is AAC or MP3 better?
AAC (M4A) is more efficient at a given bitrate, so 128 kbps AAC sounds roughly like 160 kbps MP3. Choose AAC when your playback device supports it and MP3 when you need the widest compatibility.
Related tools
Extract Audio from Video
Rip the audio track of a video to MP3, WAV or FLAC.
Video to GIF
Turn a video clip into an animated GIF.
GIF to MP4
Convert an animated GIF into a small MP4 video.
Audio Trimmer
Cut a section out of an audio file, with fades.
Chord Identifier
Name any chord from notes you pick or type.
Euclidean Rhythm Generator
Generate world-music rhythms from two numbers.