Hello,
I’m a student and hoping to do a project on building a voice imitation application. What I have in mind is like, input a recording of someone’s voice (he speaks some sentence) to the application and the application’s able to output a different sentence in a voice similar to the speaker’s. Maybe I can use a default recording and change its pitch etc. to match the speaker’s voice.
Will it be possible to build this application using Audacity? I’m very new to audio programming (as well as Audacity) with only a bit of C++ programming background. Can u pls give me an idea beforehand whether this is a realistic project so I can start working on it? 
you might be able to do it with audacity and enough programming
but a program that worked directly with synthesis might be easier
is this practical —
[sure - see it all the time on every spy show
maybe csi miami too - easier than the magic they use every monday]
— for the cia or kgb
suggest you search this forum for similar posts and responses similar to yours
then
google for voice synthesis - lots of research on that topic
doing what you want is extremely complicated and not nearly as simple as the tv or movies would make you think
as a student – cut down what you are trying to accomplish and you might win the science fair
but fooling the fbi when some pc calls in for the ransom with a voice trying to frame someone esle for a kidnapping is not feasible.
Hello,
Thanks a lot for the information.
I have researched on my topic and found out that voice synthesis is no easy task. Nevertheless, I want to give it a try.
can u suggest if there’s a more suitable open source program(running on Windows) other than Audacity which i can use?
(I’d like to use Audacity and develop this as a plug-in but not sure about it’s limitations)
Also I want to know if voice imitation will be possible by identifying the characteristics (pitch, loudness,speed…) of one recording and applying to target recording (which can be done in Audacity i think) or will it involve a lot more processing than that?
since you will do most of the programming and
audacity or whatever will do little more than i/0 for you
(record the voice sound, playback your wav after processing)
with perhaps some batch not realtime f/x
you could use audacity
but probably need a lot more f/x than audacity has
if you want realtime f/x instead of a sequence of batch processing
you really need a commercial program that has synthesis integrated with it
there are soem voice synthesis programs you might start with
not sure if they are free or not
and the quality varies a lot
google is your friend
also search in this forum for similar q&a for more pointers
and at least 2 people trying to do somethign similar
unless whoever has 3 usernames and keeps coming back to ask the same thing.
there is a lot more to synthesizing a voice than you suggest
you really need to check out the several unis that are doing research
as well as reading up on formants (wikipedia is a good start)
you will be writing a lot of code to come up with anything useful
Audacity is probably not a suitable basis for this project. Audacity is an audio editor and is not designed for real-time processing.
Also, I think this proposal is likely to be extremely complex and require considerably expertise in the field of real-time audio programming and speech synthesis.
One possible open source framework for building a speech synthesis system would be “Festival” http://www.cstr.ed.ac.uk/projects/festival/
Identifying characteristics of an individuals voice is far more complex than pitch, loudness, speed …
Here’s some links that may help give some insight into the subject:
http://www.speech.cs.cmu.edu/
http://www.speech.kth.se/wavesurfer/