# Convert Audio Bit Depth Convert audio between 8, 16, 24-bit and 32-bit float PCM with proper dithering on down-conversion. Runs in your browser, nothing is uploaded, free. ## Run it - **CLI:** `gizza tool audio-bit-depth-converter 'url=https://example.com/input' 'bit_depth=16' 'dither=triangular' 'format=wav' 'keep_metadata=true'` - **Web:** https://gizza.ai/tools/audio-bit-depth-converter/ - **Agents:** machine-readable descriptor (parameters JSON Schema) at https://gizza.ai/tools/audio-bit-depth-converter/tool.json ## Inputs - `audio` — Audio _(file; accept: audio/*)_ - `bit_depth` — Target bit depth _(field)_ - `dither` — Dither (on down-conversion) _(field)_ - `format` — Output format _(field)_ - `keep_metadata` — Keep title/artist tags _(field)_ ## Output - Converted audio (audio) ## Query parameters Open the tool pre-filled and auto-run via URL: - `bit_depth` — Target bit depth - `dither` — Dither (on down-conversion) - `format` — Output format - `keep_metadata` — Keep title/artist tags - `url` — fetch the input file from a public URL (CORS-permitting) Example: `https://gizza.ai/tools/audio-bit-depth-converter/?url=https://example.com/input&bit_depth=16&dither=triangular&format=wav&keep_metadata=true` --- ## Change audio bit depth in your browser Bit depth is how many bits each PCM sample uses — the *word length* of your audio. It sets the dynamic range: roughly 6 dB per bit, so 8-bit gives about 48 dB, 16-bit about 96 dB (the CD and streaming standard) and 24-bit about 144 dB (the studio standard). This tool requantizes an audio file to the depth you pick and writes lossless **WAV** or **FLAC**. Everything runs locally in your browser tab via ffmpeg compiled to WebAssembly, so your audio is never uploaded to a server. Anything ffmpeg can decode works as input: wav, flac, mp3, m4a/aac, ogg, opus and more. Going **down** in depth throws away low-order bits. Doing that by plain truncation turns the discarded remainder into quantization *distortion* — correlated with the signal, and most audible in fades, reverb tails and quiet passages. **Dither** fixes it: a tiny amount of noise is added before the truncation so the error becomes uncorrelated hiss instead. That is why the dither control defaults to **triangular (TPDF)** rather than off. ### Worked example You have a 24-bit master, `mixdown.wav`, and the distributor wants 16-bit. Upload the file, leave **Target bit depth** at `16`, leave **Dither** at `triangular`, keep **Output format** at `wav`, and you get `mixdown-16bit.wav` — the same length and sample rate, 16-bit words, dithered. The file is about a third smaller (24 bits per sample → 16). Want the noise tucked further out of the ear's most sensitive band? Switch **Dither** to `shibata` and re-run. ### Target bit depths - **8-bit** — unsigned PCM, ~48 dB of range. Tiny files, obvious hiss; useful for lo-fi, retro game audio and toy hardware. WAV only. - **16-bit** — the CD, streaming and delivery standard, ~96 dB. The default. - **24-bit** — studio and mastering standard, ~144 dB. Keeps headroom for further processing. - **32-bit float** — DAW interchange. Values are stored as floats, so levels above 0 dBFS survive round-trips instead of clipping. Nothing is truncated, so dither does not apply. WAV only. ### Dither algorithms - **None** — plain truncation. Correct only when you are going *up* in depth, or when another stage will dither later. - **Rectangular (RPDF)** — the minimal flat dither; lower noise than TPDF but leaves some noise modulation. - **Triangular (TPDF)** — the textbook default: fully decorrelates the error at the cost of ~3 dB more noise. Pick this if you are unsure. - **Triangular high-pass** — TPDF with the noise pushed upward in frequency. - **Lipshitz, F-weighted, Modified E-weighted, Improved E-weighted, Shibata, Low Shibata, High Shibata** — noise-shaped dithers. Same total noise energy, but moved to where the ear is least sensitive, so the *perceived* noise floor drops. `shibata` is the usual choice for a 16-bit master; `low_shibata` is the gentler variant. ### Limits and edge cases - Input files up to 10 MiB; the output is capped at 10 MiB too. - **FLAC stores 16-bit and 24-bit only.** Asking for 8-bit or 32-bit float FLAC fails with an explicit message — use WAV for those depths. - Dither is applied only when the target is an integer depth. With `bit_depth = 32f` the dither choice is ignored, because floats truncate nothing. - Increasing depth (16 → 24) does not add detail the recording never captured; it just stores the same values in wider words. - The **sample rate is left untouched** — that's a separate axis. Use the audio-resampler tool to change Hz, and audio-convert to change container or bitrate. - Embedded album art is dropped: cover images ride along as a video stream, which audio-only muxers like wav can't carry. Text tags are copied unless you clear **Keep title/artist tags**. - The output keeps the original filename with a depth suffix and the new extension (`mixdown.wav` → `mixdown-16bit.wav`, `take.wav` → `take-24bit.flac`). ## FAQ
Do I need dither when converting 24-bit to 16-bit? Yes, in almost every case. Dropping from 24 to 16 bits discards eight bits per sample; without dither the rounding error tracks the signal and shows up as gritty distortion on fades and quiet passages rather than as neutral noise. TPDF (`triangular`, the default here) is the standard, safe answer. The one time to choose `none` is when the material is already 16-bit or will be dithered by a later stage — dithering twice just stacks noise.
Which dither should I pick — triangular or shibata? `triangular` is the neutral default and is never wrong. Noise-shaped options like `shibata` keep the same total noise energy but move it into frequency bands where hearing is less sensitive, so the master sounds quieter even though a meter reads the same. Use `shibata` for a final 16-bit music master, `low_shibata` if the shaped hiss feels too bright, and stay on `triangular` for speech, archival transfers or anything that will be processed again.
Does converting to 24-bit or 32-bit float improve the sound? No. Going up in depth stores the exact same sample values in wider words — it cannot recover detail the original recording never had. It is genuinely useful before further processing, where the extra headroom and precision stop successive edits from accumulating rounding error, and 32-bit float in particular can hold levels above 0 dBFS without clipping.
Why can't I export 8-bit or 32-bit float FLAC? The FLAC format itself only defines integer sample sizes up to 24 bits in practice, and the encoder here accepts 16-bit and 24-bit. Requesting 8-bit or 32-bit float FLAC returns an error naming the supported combinations instead of silently writing something else. Choose WAV, which carries all four depths.
Does this change the sample rate or the file's length? No. Only the word length of each sample changes; the sample rate (44.1 kHz, 48 kHz, …), the duration and the channel count all pass through untouched. To change the rate, use the separate audio-resampler tool; to change container or bitrate, use audio-convert.
Is my audio uploaded anywhere? No. The page downloads an ffmpeg WebAssembly build once and then processes your file entirely inside the browser tab — the audio never leaves your device, and there is no account, queue or server-side storage.
## Related tools - [Convert an Audio File](https://gizza.ai/tools/audio-convert/): Convert any audio file to MP3, WAV, OGG, FLAC or M4A right in your browser — pick a bitrate for lossy formats. Runs locally, nothing is uploaded, free. - [Convert AIFF to FLAC](https://gizza.ai/tools/aiff-to-flac/): Convert AIFF or AIF audio to lossless FLAC in your browser — preserve samples and textual tags, choose FLAC compression level 0-12, and download the result. - [Convert AIFF to WAV](https://gizza.ai/tools/aiff-to-wav/): Free AIFF to WAV converter that runs in your browser. Choose the PCM bit depth, sample rate and channel layout — at a matching depth the samples are identical. - [Strip Audio Metadata](https://gizza.ai/tools/audio-metadata-stripper/): Remove ID3 tags, comments, chapters, and cover art from MP3, FLAC, OGG, M4A, WAV, and other audio files in your browser. Stream-copy audio, no upload. - [Extract Audio from a Video](https://gizza.ai/tools/extract-audio-from-video/): Pull the audio track out of any video as MP3 or WAV, right in your browser — pick a format and bitrate. Runs locally with ffmpeg, nothing is uploaded, free.