How does Change Opus Speed change the playback rate?
+
Concretely, the tempo is changed with a WSOLA time-stretch that leaves pitch alone, so speech stays intelligible instead of turning into a chipmunk; this needs a re-encode, so choose your speed once. Choose a multiplier — 0.5x for half speed, 2x for double — and the whole file is retimed.
Will speeding it up make voices sound high-pitched?
+
No. The time-stretch preserves pitch, so a 1.5x speech track sounds like the same person talking faster rather than like a chipmunk. That is the difference between a proper time-stretch and simply playing the samples faster.
Does the source format affect the result?
+
It does. Opus is the best-in-class lossy codec at low bitrates and the standard for voice and real-time audio. That determines whether the retimed result can be written back cheaply or has to be fully re-encoded.
Anything specific to Opus here?
+
Yes — Opus always resamples to 48 kHz internally, so sample-rate choices you make elsewhere in the chain do not survive into it. Worth knowing before you commit to a large speed change.
How far can I push the speed before it sounds wrong?
+
Roughly 0.75x to 2x is transparent for speech. Beyond about 2.5x the stretch artefacts become audible on any algorithm, and below 0.5x the reconstruction has to invent too much. Music tolerates far less than speech does.
Does the file get shorter, and does it get smaller?
+
Shorter, yes, and in exact proportion — an hour of audio at 1.5x is forty minutes. The file shrinks by roughly the same proportion because there is less material to encode. The bitrate, the sample rate and the channel layout are all left exactly as they were.
