I am trying to edit the tone and pitch of a recording of my voice so that it comes close to imitating another person’s voice.
I have a recording of that other person’s voice in Audacity, and am wondering what is the best way to edit mine to try and mimic/match the other person. I have adjusted pitch, and that made a little difference, but the tone qualities are so different that it still sounds nowhere near the other person. Does anyone have a suggestion for how I would best approach doing this in Audacity?
Voice conversion, when it’s done at all, is done by very specialized software – usually money-based. There is no shortage of requests to turn an X voice into Y. X and Y can be Man, Woman, Child, Old Person, and Announcer (that’s a popular one).
Pitch shifting never works very well past one or two piano notes up or down. Past that, the result is very unnatural partly because only a portion of the voice must be shifted – the basic tones, but not the sibilants or plosives. Pitch shifting moves everything.
Also, as any mimic or comedian will tell you, voice conversion must be approached with rhythm and style, not just a pitch change. If you have a flat, monotone speech like most people, the announcer software will turn you into a flat monotone announcer.
You can also mess with the equalizer and vocoder. All this is in Audacity 1.3.12. The earlier Audacity versions didn’t have good enough tools.
I guess that was me thinking firmly inside the box. You don’t have to convert the voice. You can start with printed text and then, we assume, patch it up a little here and there if the automatic synth software misses it.
I guess if you were really into pain and torture, you could use capture software to print your words, and then the synth software to generate a new voice.
I know the first inclination is to chuckle, but if you wanted to do this job using existing software…