Audio Speed Changer
Change audio speed from 0.5× to 2×, with either tape-style pitch movement or pitch-preserving tempo adjustment. The local preview is responsive, while export is authoritative and uses FFmpeg resampling or atempo according to your setting. Trim and preview before rendering MP3, WAV, FLAC, M4A, or OGG. Optional AI and transcription actions use separate server jobs.
Core edit stays local
AI and transcript are opt-in server steps
Drop an audio file here
or click to browse
Use this when
- You need a slower or faster practice, speech, or music file.
- You can choose between tape-style pitch movement and pitch-preserving tempo.
Choose another approach when
- You need sample-identical output or a forensic transformation.
- Artifacts from extreme time stretching would be unacceptable.
Reproducible example
Listen to the CLI reference fixture and output
These deterministic files are FFmpeg CLI references, not measurements of the shipped browser renderer or microphone capture. Automated checks are shown below; independent human review remains pending.
Synthetic input
SHA-256 aebecf107510c32ee13e6c58e1b56edaaed84cc753c3af4dea8163f37eb0aa18
- Duration: 2.4 s
- Size: 423.4 kB
- Sample rate: 44,100 Hz
- Channels: 2
Rendered output
SHA-256 64a545d854b433d2e2ec33f837e431ceac8170dcb450154e60d7a5be7528b8ca
- Duration: 3.193 s
- Size: 563.4 kB
- Sample rate: 44,100 Hz
- Channels: 2
Automated fixture check
- Output duration Pass.
Observed: 3.1934 s. Expected: 3.2 ± 0.04 s. - Preserved reference frequency Pass.
Observed: 523.2 Hz. Expected: 523.25 ± 1 Hz.
Environment
audio-v1 · Sep 3, 2026, 2:44 AM UTC
FFmpeg CLI · darwin arm64 · v23.7.0
ffmpeg version 8.1.2 Copyright (c) 2000-2026 the FFmpeg developers
ffprobe version 8.1.2 Copyright (c) 2007-2026 the FFmpeg developers
@ffmpeg/ffmpeg 0.12.15 with @ffmpeg/core 0.12.10
- Public evidence manifest
- evidence.json
- Fixture license
- CC0 1.0
- Generator source
- generate-audio-flagship-evidence.mjs.txtSHA-256 228ca719f843debbddd1e1b3482f2b6d406b3d89afab34ef7797eacf01c998c7
- Source base commit
- e11eef48f52df732e0ef32d69f7ea10635b150bc
- Browser renderer source
- src/lib/audio/editor.tsSHA-256 159c26bb4f175dc8c823d2c70178f631d0125cf8059b3f8ea1494a3f3fba6479
The repository is private. The exact generator snapshot is public; the browser renderer and commit are retained privately and pinned here by full hash.
Fixture rights: CC0 synthetic audio generated in the repository. No customer or third-party recording is used.
Reference command
ffmpeg -y -hide_banner -loglevel error -i public/fixtures/audio/v1/speed-metronome-523hz.wav -filter:a atempo=0.7500 -vn -c:a pcm_s16le -map_metadata -1 -bitexact public/fixtures/audio/v1/speed-075-keep-pitch-reference.wavLimitations and failure modes
- The live Keep pitch preview is approximate; the FFmpeg export is authoritative.
- Both speed modes create a new render. Quality depends on the source and the selected factor.
Methodology
The public corpus is deterministic. Full hashes, commands, engine version, and semantic signal checks are recorded in the manifest. These files exercise the native FFmpeg reference graph; browser workflow behavior is a separate release gate and device speed is not inferred.
Technical sources
How it works
Load your audio
Choose a supported audio file. It is decoded on this device and appears on the timeline.
Choose speed and pitch
Set 0.5× to 2×. Keep pitch uses FFmpeg atempo on export; tape style moves pitch with speed.
Preview and export
Compare the preview, then render the timeline locally in the format you need.
Questions from current search and product use
Can I change speed without changing pitch?
Yes. Enable Keep pitch and export uses FFmpeg atempo. The live preview is an approximation, so the exported file is the authoritative result.
Can I slow music down for practice?
Yes. A setting below 1× with Keep pitch enabled is useful for practice, lectures, and spoken audio. Results vary with the source and more extreme settings can introduce artifacts.
Does speed changing preserve every sample?
No. Both modes create a new render. Tape style resamples the signal, while Keep pitch uses time stretching. Export WAV or FLAC to avoid adding a lossy output codec, not to make the transformation lossless.
Is my audio uploaded?
Core editing, preview, and export stay in your browser. Optional AI Enhance and Transcript actions are separate server jobs with an explicit disclosure.
Accountability
- Author
- Jesse Delia
- Technical review
- Pending independent review
- Report a correction
- support@filemorf.com
Change history
2026-07-30: Tool-first page, precise processing boundary, synthetic fixture evidence, limitations, and accountability record added.