How I change the words and keep the voice in a speech !

Im looking for a software wich I can edit the speech : change the words and keep the voice

no one knows it :P ?

sorry, i don't understand exactly what you mean.

please tell us more.

Do you mean text-to-speech type software or software for editing out certain words from an audio file?

What you're asking is nearly impossible. You want to use some kind of voice synthesizer to exactly duplicate someone else's voice timbre, intonation, pronounciation, inflection, etc. etc. and you expect to generate a spoken phrase with a realistic voice that's indistinguishable from actual human speech.

The latter has already been done, there are frighteningly good speech synthesizers, but exactly adjusting the voice beyond a generic male/female sound is something highly specialized and extremely difficult. Human voices are amazingly complex waveforms.

I doubt you'll really find any software for Windows to do this, other than forensic tools that aren't available to consumers and cost thousands of dollars. If they even work without custom hardware.

yeah I want to take a speech of someone and edit the words but keep the same voice , I found some modified speeches but I dont know if the normal consumer did it or not

The only way you can really do this without having access to FBI resources is to use audio editing software (Audacity is a very good free sound editor) to chop up a spoken sample and rearrange the bits to form a new sentence.

Of course you'll be limited to whatever spoken words there are and you can't change the way they're spoken...at most you could do some trickery with quick pitch changes and the like, but the result will be far from natural.

Consider this...time-stretching is a very popular sound processing technique and basically means altering the speed without altering the pitch. This is typically done by chopping the sound into tiny bits and repeating some of them so as to artificially lenghten the sound (or skipping bits to shorten the output) but keep the pitch. There are decent algorithms to do this with very little artifacting.

There's one way to do a perfect timestretching, but it's computationally incredibly intensive. Basically it means "disassembling" the voice into its component sine waves - every sound, no matter how complex, is ultimatively just an addition of sine waves of all kinds of frequencies and amplitudes - and then resynthesizing the sine waves with different time lengths. This process is known as FFT (Fast Fourier Transformation).

A few years ago, some geeks processed a sample of a spoken phrase in French that lasted 1.5 seconds. It took the computer several hours to create a perfect timestretched copy.

Guys ,do you know any good free text-to-speech software ?
is there software can remove the speech from the songs ?

baxxter said:

is there software can remove the speech from the songs ?

Most recorded songs are in stereo with the vocal in both the left and right channel.

You will have little success in removing the vocal portion in such a scenario. Consumer software is going to react to the vocal and instruments based on their frequency and cannot delete a 2000 Hz passage of human voice without also removing the 2000 Hz guitar notes that are in the same portion of the song.

You can fiddle with an equalizer to have some effect, but don't expect much.

If the vocal track happens to be ONLY in one channel and has nothing else in that channel, then you could of course delete it, but that is rarely the case.

baxxter said:

Guys ,do you know any good free text-to-speech software ?

Free, not that I'm aware of...but you could try this website (you can download each portion), it might do for what you want:
AT&T Natural Voices® Text-to-Speech Demo

baxxter said:

is there software can remove the speech from the songs ?

It's practically impossible to get a vocal completely removed from a song without losing other details in various instruments.

*IF* you have an instrumental only version of the exact same song, you can subtract one from the other to get a full Vocal only version. (just throwing that out there, but it's not what your asking)

The vocals are pretty much always placed in the center of the mix. (although occupying various parts of the stereo field as ignatzatsonic mentioned). You can use Audacity (free) and it has a few ways to remove most of the vocals (but again, the overall final sound will not be great)

Plugin method:



Split-mono method:





Hope that helps a little.