Change Audio Speed (Keep Pitch)

Change audio speed in your browser

Pick an audio file and a speed factor: values above 1 play it faster, values below 1 play it slower, and the pitch stays exactly where it was. This is a time stretch (also called a tempo or BPM change), the opposite of the tape-speed effect — a slowed track keeps its key instead of dropping in pitch, and a sped-up podcast keeps its normal voices instead of turning to chipmunks. Use it to listen to podcasts and audiobooks at 1.5×, slow a solo or a language lesson down for transcription and practice, or nudge a track to a target BPM. Drag the slider, type a decimal, or hit one of the preset chips. Everything runs entirely in your browser with ffmpeg compiled to WebAssembly, so your audio is never uploaded to a server.

Worked example

Speed up a lecture recording to play half again as fast: upload the file, set Speed factor to 1.5, leave Format on mp3, and run. The result plays 33% shorter at the same pitch and downloads as the original name with a -time-stretched.mp3 suffix. Some more common factors:

• 2 — double speed (half the duration); a 60-minute podcast becomes 30 minutes.
• 1.25 — a gentle podcast/audiobook speed-up that stays easy to follow.
• 0.75 — 25% slower, a common setting for transcribing speech or learning a fast musical passage.
• 0.5 — half speed (double the duration), slow enough to pick out every note in a solo.

Factor, percentage, and BPM

The factor is a plain multiplier of playback speed, so it converts directly to the other ways speed gets described:

• Percentage — multiply by 100. 1.5 is 150% speed, 0.75 is 75% speed. A "20% faster" request is 1.2; "40% slower" is 0.6.
• BPM — the factor is the ratio of the two tempos: factor = target BPM ÷ source BPM. To take a 90 BPM loop to 120 BPM, set 120 / 90 ≈ 1.33. To take a 128 BPM track down to 100 BPM, set 100 / 128 ≈ 0.78.

How it works

The tool feeds the audio through ffmpeg's atempo filter, a WSOLA time-stretch that changes tempo while leaving pitch untouched. atempo accepts 0.5× to 2× per pass, so factors outside that window are split into a chain whose product is your factor (for example 4× becomes 2× then 2×, and 0.25× becomes 0.5× then 0.5×). Any attached album-art image is dropped so audio-only formats like WAV encode cleanly, and the result is re-encoded to the format you pick.

Formats

• MP3 — lossy at 192 kbps; small and playable everywhere (the default).
• WAV — lossless 16-bit PCM; largest, ideal for further editing.
• OGG — lossy Vorbis at 192 kbps; open format, good quality per byte.
• FLAC — lossless and compressed; smaller than WAV, still a perfect copy.
• M4A — AAC in an mp4 container; good quality at small sizes.

Limits and edge cases

• Input files up to 10 MiB; anything ffmpeg can decode works (mp3, wav, flac, m4a/aac, ogg, opus, and most video containers' audio tracks).
• The factor ranges from 0.25× to 4×. Extreme stretches sound increasingly processed — the WSOLA artifacts (a slight fluttery or metallic quality) grow as you move away from 1; between about 0.7× and 1.5× they are hard to hear on most material.
• 1 is rejected as a no-op rather than silently re-encoding your file — use the audio-convert tool if you only want a format change.
• The output is always re-encoded (the filter chain requires it).
• This is a pure time stretch: it changes tempo only. To change the key without changing the speed, use the audio-pitch-shift tool; for the tape-speed effect where pitch and speed move together, use the change-speed tool.

FAQ