Change Audio Speed

How the Audio Speed And Pitch works

Change audio speed: what to know first

There are two genuinely different things people mean by changing the speed of audio. The first is resampling: play the same samples out at a different rate, and the recording gets shorter and higher, or longer and lower, together. The second is time stretching: keep the pitch where it is and change only the duration, which requires re-synthesising the signal with a phase vocoder and always leaves some artefact behind. This tool does the first, deliberately and cleanly, and tells you so rather than pretending otherwise.

Why tape-style is the right default for practice. If you are working a fast passage out by ear, dropping to 70 percent speed makes every note audible, and the fact that it also drops a little over five semitones in pitch is usually irrelevant — you are listening for articulation and order, not absolute pitch. The sound stays completely natural because nothing has been invented: every sample is the original sample, just clocked differently. Time-stretching the same passage would keep the key but introduce the smeared, slightly metallic quality that phase vocoders produce on transients, which is precisely the detail you were trying to hear.

Transposing in semitones. Switch the mode and the semitone field takes over, converting your request into the matching playback rate: one semitone up is a rate of about 1.059, an octave up is exactly 2.0, an octave down exactly 0.5. This is the practical way to move a backing track into a key that suits a voice, or to match a recording made at a different reference pitch. The trade-off is the duration: transposing up by two semitones also makes the track about 11 percent shorter, so it will no longer line up with a click or a video.

What the readout tells you. You always see both consequences of whatever you asked for — the playback rate, the pitch change in semitones, and the resulting length next to the original. That symmetry is the point: there is one control here wearing two hats, and seeing both numbers stops you being surprised by the one you were not thinking about.

The export is a real re-render. Rather than recording the playback, the file is re-rendered in an offline audio context at the new rate, at the original sample rate and channel count, and written out as 16-bit PCM. That means the export is faster than real time and is bit-for-bit repeatable, and it means a four times slowdown produces a file four times longer with no dropouts, which is not something you can rely on from a screen recording.

Practical limits. The range runs from a quarter speed to four times. Below about half speed, resampling artefacts and the original recording's own noise floor start to become obvious on anything compressed, and above about double speed you will hear the top end fold away as content moves past what the sample rate can represent. Both are properties of the source material rather than bugs, and both are why extreme settings are best used for diagnosis rather than listening.

Related tools