I use Asterisk telephony system and need to create customised messages (sounds in Asterisk jargon) to be plaid during a phone call.
I can record + convert the messages with no problem - the fact is I have a strong male voice, whereas a female warm voice would be more apt for the job.
Short of paying some professional to do the job, is there a way with Audacity to convert my own voice recording to something as close as possible to a natural female warm voice that doesn’t sound artefact? Is there a guidelines documentation to help me find the right setting in order to achieve my objective?
I’ve tried changing the pitch but it sounds unnatural.
Yes, that’s exactly the problem.
A small change to pitch can work quite well, but large changes will sound unnatural for a couple of reasons:
Stretching audio is a “lossy” process - there’s always some loss of sound quality because the process involves chopping up the audio into little pieces, squashing them together (or spreading them out), then sticking all the pieces back together again.
The difference between male and female voices is not only the pitch, but also involves resonance, harmonic content and other factors. Some sounds (for example most consonants) will sound very similar for either a male or female voice, where as other sounds (for example vowel sounds) will be significantly different. Changing the pitch of the audio will apply the same pitch shift to all sounds, regardless of context.
So basically, the problem is far more complex and difficult to achieve.
Probably the easiest solution is to buy a nice box of chocolates for your mother/sister/daughter/wife/girlfriend/girl in the office, and ask them nicely if they could help you out.
Short of going bankrupt buying chocolates for all my girlfriends until they agree to cooperate, is there any open source voice synthesizer worth mentioning?
“TextAloud” is a voice synthesizer for Windows that supports a huge range of 3rd part voices, including “AT&T Natural Voices”.
The basic price is about $30, but additional high quality voices are extra.
There’s a load of demo samples here: http://www.nextup.com/index.html
Even the best voices sound computer generated, and I expect it would be more fun to go bankrupt buying chocolates for all of your girlfriends
I’ve found DSPEECH here (http://dimio.altervista.org/eng/) - Surely it’s not the ideal in terms of the voice sound natural, but… hey, at least I save the chocolates for a better occasion.
I expect it would be more fun to go bankrupt buying chocolates for all of your girlfriends
Well, that. But you only need one. Listen with your ears and pick the one woman with the right voice for the job. You have to steel yourself. This may be more difficult than a keyclick.
We have two women in the office that do a lot of the voice-over work. One is an out-of-work actor and was the subject of the last shoot I did. (OK, now try it with a breathless Cockney accent.) The other just has that smooth velvety Lauren Bacall voice. We found her on the elevator/lift between floors. “Would you press five, please.”
“Um. Have you ever done voice work?”
“People keep telling me I should.”
“This is your lucky day.”
We have some fascinating anomalies, too. One guy has a terrific voice and can’t read copy at all. Completely tone deaf.
Asterisk telephony system
Our Asterick system uses “Comedian Mail” and has a very nice woman’s voice doing all the work. We didn’t shoot it.
In the UK there is a cockney soap opera on BBC TV called “Eastenders”,
e.g. here is how they pronounce “thing”, “throat” and “breathe” … “fing , froat and breeve”