Audio Formats and Quality Explained: WAV, MP3, AAC, FLAC, and When Each One Matters

Someone asks for a "high quality" version of a podcast episode and you export a 60 MB WAV file to share on WhatsApp. "Upload failed." You drop it to 128 kbps MP3 and the file lands at 1.6 MB, but the cymbals now sound like someone crumpling aluminum foil. Both choices were wrong — one was too big, the other was too lossy. Understanding why sits squarely in the mechanics of audio formats, bit depth, sample rate, and bitrate. This guide walks through the entire chain, with real numbers, real math, and practical rules for picking the right format every time.

The foundation of digital audio: bit depth and sample rate

Sound is a continuous pressure wave. To store it digitally, a converter samples that wave thousands of times per second and records the amplitude at each point. Two numbers control quality: sample rate (how many snapshots per second) and bit depth (how precisely each snapshot is measured).

Standard CD audio uses 44.1 kHz — 44,100 samples every second — and 16-bit depth. The Nyquist theorem guarantees that a 44.1 kHz rate captures every frequency up to about 22 kHz, which is the ceiling of human hearing. Going higher (48 kHz, 96 kHz, 192 kHz) captures ultrasonic content that humans cannot hear but audio engineers may use for processing headroom. Bit depth defines the amplitude resolution: 16-bit gives 65,536 discrete levels per sample and a theoretical dynamic range of about 96 dB. 24-bit recording pushes that to 16.7 million levels and roughly 144 dB — more headroom means less clipping risk during mixing and mastering.

Bit depth and sample rate are the reason uncompressed audio is large: every single sample is written to disk with no shortcut. A 44.1 kHz, 16-bit, stereo file produces 44,100 × 2 bytes × 2 channels = 176,400 bytes per second. Multiply by 60 and you get roughly 10.1 MB per minute. That is the cost of preserving every detail of the waveform.

WAV and PCM: what raw audio looks like on disk

A WAV file is a thin RIFF container wrapping raw Pulse-Code Modulation data — the literal stream of sampled amplitudes, uncompressed and unprocessed. There is no algorithm discarding information, no frequency ceiling imposed by a model. This is why WAV is the reference standard in recording studios, broadcast, and mastering: it carries everything the microphone captured.

The raw math for a one-minute CD-quality stereo file: 44,100 samples/sec × 2 bytes (16-bit) × 2 channels × 60 seconds = 10,584,000 bytes ≈ 10.1 MB. A full hour of the same recording crosses 600 MB. AIFF is Apple's equivalent container — same size, same concept, different metadata header. The takeaway: uncompressed audio trades massive file sizes for perfect fidelity.

How lossy compression actually works: psychoacoustic masking

Lossy formats like MP3 and AAC do not "reduce quality uniformly." They exploit the limitations of human hearing through a model called psychoacoustic masking. When two sounds occur at the same time, the louder one can mask the quieter one — your auditory system simply does not perceive the masked signal. A cymbal crash buries a nearby guitar harmonic. A loud kick drum hides a quiet bass note. The codec analyzes each frame of audio, identifies what the ear will not miss, and permanently throws that data away.

The second mechanism is the frequency ceiling. Human hearing spans roughly 20 Hz to 20 kHz, and sensitivity drops steeply above about 16 kHz. Most MP3 encoders apply a hard low-pass filter that removes everything above 17–18 kHz — even at the highest bitrates. This is inaudible to most listeners over 20 or anyone listening through earbuds on a commute. But it is one reason audiophiles prefer lossless formats for critical listening: the data is gone, and no decoder can bring it back.

Bitrate: what 128, 192, and 320 kbps actually mean

Bitrate, measured in kilobits per second (kbps), controls how much data the lossy codec keeps. Lower bitrate means more aggressive filtering and discarding. The practical tiers for MP3 are well understood:

• 128 kbps — the long-time "default" quality. The encoder removes most frequencies above 16 kHz, compresses stereo image, and smears transient detail. Cymbals lose shimmer, guitar strums sound smeared. Acceptable for casual listening on low-quality speakers, but noticeably degraded on good headphones. An entire album fits on a single CD at this rate.
• 192 kbps — a meaningful jump. More high-frequency content survives, stereo width improves, and most casual listeners cannot distinguish the source from a WAV in a blind test. This is the practical minimum for music you intend to keep.
• 320 kbps — the maximum for MP3. Psychoacoustic filtering is minimal; only the most extreme high-frequency cutoffs remain. In controlled listening tests, even trained ears struggle to distinguish 320 kbps MP3 from a WAV source. For almost every listener and every use case, this is perceptually lossless.

AAC tells a slightly different story: at 128 kbps AAC, quality typically matches or exceeds 192 kbps MP3. Apple's encoders and the open-source FDK-AAC library both produce cleaner results at equivalent bitrates, which is why Apple Music and YouTube default to AAC rather than MP3.

The math: file sizes you can actually calculate

File size for lossy formats is direct multiplication: bitrate × duration. One minute of audio at 128 kbps:

• 128 kbps × 60 seconds = 7,680 kilobits = 960 KB ≈ 0.94 MB
• 192 kbps × 60 seconds = 11,520 kilobits = 1,440 KB ≈ 1.41 MB
• 320 kbps × 60 seconds = 19,200 kilobits = 2,400 KB ≈ 2.34 MB
• CD-quality WAV (44.1k/16-bit stereo) = 10,584,000 bytes ≈ 10.1 MB

The comparison is stark: 320 kbps MP3 delivers nearly identical perceptual quality to WAV at roughly one-quarter the file size. That is the entire value proposition of lossy compression. For FLAC, the ratio is different — lossless compression typically reduces WAV by about 50–60%, so a one-minute CD-quality FLAC file lands around 4–6 MB, roughly half the WAV size with zero quality loss.

Format comparison: when to use each one

Browser support and compatibility reality

Format choice is not purely a quality question — it is also a deployment question. MP3 has universal support: every browser, every operating system, every car stereo, every Smart TV. If you need guaranteed playback with no caveats, MP3 is the only format that delivers it.

AAC plays natively in Chrome, Firefox, Edge, and Safari — covering the vast majority of modern desktop and mobile users. However, some older Android devices and Linux distributions may struggle without additional codecs. FLAC playback is well-supported in Chrome and Firefox, but Safari's support is limited to macOS Ventura (13.0) and later. Users on older macOS or iOS versions will hear nothing. OGG Vorbis is the most restricted: Safari has never supported it natively, and Apple shows no signs of adding it. If your audience includes any Safari users, OGG is not a viable primary format — always pair it with an MP3 fallback.

Trimming, loudness, and practical editing considerations

When you cut audio or adjust volume, format choice directly affects precision and quality. The audio trimmer performs sample-accurate cuts on WAV files because every sample is independently accessible. Compressed formats like MP3 encode in frames of 576 or 1,152 samples, so trimming requires the encoder to reconstruct the frame, cut, and re-encode — each edit introduces a tiny quality loss that accumulates across many cuts.

Loudness is measured in LUFS (Loudness Units relative to Full Scale), and it is the standard that Spotify, YouTube, Apple Music, and most streaming platforms use for normalization. A quiet podcast episode at -24 LUFS will be boosted by the platform to roughly -14 LUFS, while a hot recording at -8 LUFS will be turned down — often with unwanted pumping artifacts. Targeting -16 LUFS for podcasts and -14 LUFS for music before uploading gives you control over the sound rather than letting an algorithm decide.

// Quick loudness estimate (simplified, not a true LUFS meter)
function estimateLUFS(samples) {
  const sumOfSquares = samples.reduce(
    (acc, s) => acc + s * s, 0
  );
  const rms = Math.sqrt(sumOfSquares / samples.length);
  // -0.691 is the ITU-R BS.1770 relative level offset
  return 20 * Math.log10(rms) - 0.691;
}

Audio tools in Convrs

All of this runs in the browser on Convrs — no audio file ever leaves your device. The WAV to MP3 converter transcodes uncompressed WAV into high-quality MP3 with adjustable bitrate, preserving bit depth and sample rate without a server roundtrip. The audio trimmer cuts, trims and rearranges audio clips directly in your browser with sample-level precision. The volume booster adjusts loudness with real-time peak monitoring, so you can target the right LUFS level before sharing or uploading. Three tools, zero uploads, complete privacy.