Understanding Audio File Formats: MP3, WAV, and More

Audio file formats divide into two families. Uncompressed formats (WAV, AIFF) store audio as raw waveform samples — a sequence of numbers representing the sound pressure at each instant. Compressed formats (MP3, AAC, Opus) apply algorithms that reduce file size by discarding or transforming information the ear tolerates losing. The distinction mirrors the lossless-versus-lossy divide in images: uncompressed is exact and large, compressed is small and lossy.

Within compressed audio, there is a further split between lossy and lossless compression. Lossy formats (MP3, AAC, Ogg Vorbis, Opus) discard audio data permanently to achieve small sizes. Lossless compressed formats (FLAC, ALAC) reduce size without discarding information — the decoded audio is bit-identical to the original — but they achieve smaller reductions than lossy formats, typically 40 to 60 percent of the original WAV size rather than the 10 percent a high-quality MP3 achieves.

The choice follows the use. For archival or editing, where quality matters and size is secondary, use WAV or a lossless compressed format like FLAC. For distribution and streaming, where size and bandwidth matter, use a lossy format. For most web audio, a 128 kbps MP3 or AAC file is indistinguishable from the source for casual listening while being a fraction of the size.

WAV (Waveform Audio File Format) is the standard uncompressed format. It stores raw PCM samples with a header describing the sample rate, bit depth, and channels. A CD-quality WAV (44.1 kHz, 16-bit, stereo) consumes about 10 MB per minute — large, but lossless and universally supported. WAV is the right working format for recording and editing, because every operation preserves the full fidelity.

MP3 (MPEG-1 Audio Layer III) is the most widely compatible lossy format. It achieves roughly a 10:1 compression ratio at 128 kbps with artifacts that are inaudible to most listeners in most conditions. Its universal support — every device, every browser, every player — is its enduring advantage, even though newer formats achieve better quality at the same bitrate.

AAC (Advanced Audio Coding) is the modern successor to MP3, achieving better quality than MP3 at the same bitrate. It is the default audio format for YouTube, Apple Music, and most streaming platforms. Opus is the newest lossy format, with excellent quality at very low bitrates and growing browser support, making it attractive for real-time communication and streaming. FLAC is the standard lossless compressed format, halving WAV size with no quality loss, favored by audiophiles and for archival.

Sample rate is how many times per second the audio waveform is measured, measured in Hertz. CD audio uses 44.1 kHz (44,100 samples per second), which captures frequencies up to about 20 kHz — the upper limit of human hearing. Higher sample rates (48 kHz, 96 kHz) capture frequencies above human hearing, which matters for some production workflows but not for final distribution, because no listener can hear the difference.

Bit depth is how many bits represent each sample, determining the dynamic range — the difference between the quietest and loudest sound the format can represent. 16-bit (CD quality) provides about 96 dB of dynamic range, far more than any listening environment uses. 24-bit provides more headroom for recording and editing, where processing can introduce noise that the extra range absorbs, but it offers no audible benefit for final distribution.

The practical takeaway: for distribution, 44.1 kHz and 16-bit (CD quality) is more than sufficient, and higher settings only inflate file size. For recording and editing, higher sample rates and bit depths give you processing headroom that you will downsample to 44.1/16 before distribution anyway. Matching the format to the stage of the workflow prevents both quality loss and wasted size.

Bitrate, measured in kilobits per second (kbps), is the primary determinant of lossy audio quality and size. Higher bitrates preserve more of the original audio but produce larger files. The relationship is roughly linear: a 256 kbps file is about twice the size of a 128 kbps file of the same duration.

The useful ranges are well established. 64 kbps is acceptable for speech where size matters most (podcasts distributed to constrained audiences). 128 kbps is the standard for music distribution — indistinguishable from the source for most listeners in most conditions. 192 to 256 kbps is high quality, suitable for discerning listeners. 320 kbps is the maximum for MP3 and is essentially transparent — no audible difference from the original for any listener in any condition.

Variable bitrate (VBR) adjusts the bitrate moment to moment based on the complexity of the audio, allocating more bits to complex passages and fewer to silence or simple sections. VBR achieves better quality at a smaller average size than constant bitrate (CBR), at the cost of less predictable file sizes. For most distribution, VBR is the better choice; for streaming where exact size matters, CBR predictability wins.

Before encoding to a lossy format, trim the audio to the needed section. Removing silence at the start and end, cutting unwanted sections from the middle, and trimming to the relevant excerpt reduces duration, which reduces file size directly — every second removed is a second not encoded. The Audio Tools include a trimmer that performs these cuts losslessly on the source before you encode.

Normalization adjusts the overall volume so the audio peaks at a target level, making quiet recordings audible and preventing loud ones from clipping. It is a global adjustment — it scales the entire waveform uniformly — and it does not compress dynamic range the way a compressor does. Normalize before encoding so the lossy encoder works with a well-leveled source, which produces cleaner results at a given bitrate.

Merging multiple clips into a single file is another pre-encoding step. Combining separate recordings (intro, main content, outro) into one file before encoding avoids the quality loss of encoding each separately and then concatenating, which would re-compress the already-compressed audio. The workflow — trim, normalize, merge on the lossless source, then encode once to the final lossy format — preserves maximum quality through every step.

Consider a 45-minute podcast recorded at 44.1 kHz, 16-bit. As an uncompressed WAV, it weighs roughly 450 MB — far too large to distribute. The content is speech, not music, which changes the bitrate calculus: speech has less complex frequency content than music, so it tolerates lower bitrates without audible degradation.

Encoding to MP3 at 128 kbps (the music standard) produces a file of roughly 42 MB — 45 minutes times 60 seconds times 128 kilobits, divided by 8 for bytes, gives about 43 MB. But for speech, 64 kbps is widely accepted as transparent: listeners cannot distinguish it from the source on phone speakers and earbuds, and the file drops to roughly 21 MB. For a podcast distributed to a mobile audience where data matters, that halving is a meaningful win with no perceptual cost.

The decision framework: match the bitrate to the content, not to a default. Music demands 128 kbps minimum, 192 to 256 for discerning listeners, 320 for archival transparency. Speech tolerates 64 kbps, sometimes 48 kbps for talk-only content. Variable bitrate squeezes further by spending bits only where the audio is complex. The size scales linearly with bitrate — halve the bitrate, halve the file — so the question is always: what is the lowest bitrate my audience cannot distinguish from the source on their actual devices? Verify by listening, not by trusting the number.

After encoding, listen to the result on the devices your audience will use. Artifacts that are inaudible on studio headphones can be noticeable on cheap earbuds or phone speakers, which emphasize different frequencies. A quick check on a phone speaker catches the pre-echo and swishy high-frequency artifacts that are the hallmarks of aggressive MP3 compression.

Keep the lossless source. Encoding to a lossy format is a one-way operation — the discarded audio data cannot be recovered from the MP3. If you later need a higher quality version or a different format, you must re-encode from the lossless original, not from the lossy file, because re-encoding a lossy file compounds artifacts. The Audio Tools process your files locally in the browser, so the source never leaves your device during trimming, merging, and normalizing — but the encoding decision itself is still one-way, and keeping the original is the discipline that protects you.

Finally, match the format to the audience constraints. MP3 for maximum compatibility, AAC or Opus for better quality at the same size, FLAC for archival and audiophile distribution. The right choice is the smallest format that is transparent to your listeners on their devices — and verifying that transparency by listening, not by trusting a bitrate number, is what makes the choice reliable.