I’m looking for ways or tools to change the pitch of audio without messing up the sound. Ideally, it should work for vocals or full tracks without adding noticeable noise or distortion.
Have you tried using Alt+up/down arrows? That uses a a novel state-of-the-art algorithm developed by Staffpad and open-sourced by Audacity
If you click the 3-dot menu, you can also go to Pitch and Speed and enable “optimize for voice” to get something nicer.
Theoretically it should be possible to do this without any quality loss, no?
only if you also slow it down. The pitch of a sound is how often the sound wave changes (frequency), so if you slow it down, the frequency decreases, too. In a physics sense, this is the only way to preserve the waveform perfectly. In a psychoacoustics sense, formant preservation is probably superior as it doesn’t unnaturally slow down the voice and/or cause a chipmunk effect
If you change speed and pitch together (with the Change Speed effect) it can be perfect with speed change as the only side-effect.
But if you change it too much it won’t sound perfect because there are vocal sounds that don’t change pitch when you sing a different note. You can get similar unnatural results with instruments but it’s usually most-noticeable with voice.
Changing pitch or speed independently is VERY COMPLICATED. It was impossible in the analog days. It uses FFT to convert from the normal time domain (waveform) to the frequency domain, then the frequencies can be shifted and reverse FFT is applied. FFT is “imperfect” and requires compromises when you have regular constantly-changing audio.
AI can probably to it if you can find the right AI tool because it can completely create or re-create audio.
OK, I should have said “perceptible” or “significant” quality loss … and the first thing we’d have to do is define quality, and the significance level or perception threshold 