Changing Recorded Voice other than Pitch?

[thudo]

They are NOT inexpensive but there is variety. There is no reason for software in 2010 to do a lot more these days especially with powerful PCs and 64bit (where possible).

[donrandall]

Make a copy of the original. File the original safely away. Work with the copy. Feel free to disregard this advice if you so choose. You will then, at some point, be thrilled by the joy of discovery when you come to understand just exactly why this is really very good advice.

You can buy some of the available "voice changing" or "voice morphing" software and play around and be disappointed by what your money buys if you like. But why spend fifty or a hundred dollars or so to get something that won't do anything you can't already do with Goldwave? Goldwave can do some really good things, but it is the tool and it is only capable of giving us what we are willing to work for.

Stop and think about it: For the most part, what they (those other software programs) are doing is raising and lowering pitch by changing speed. They can also maintain the new pitch when restoring the original timing of the audio ( which was altered when speeding up or slowing down to change pitch).

They can also play around with equalizing and compression to further distort the sound. Make no mistake: What we are talking about is a distortion of the original. Not all distortion is bad. Distortion can be a good thing if it helps achieve the desired result.

But with inexpensive software, the best we can reasonably expect is a cartoonish pretense of converting a man's voice to a woman's voice or any of the other transitions they claim to offer.

With a little bit of thought and experimentation, one can achieve pretty much the same results using Goldwave. Possibly better results.

Example: Highlight a bit of a track containing a few seconds of voice, note the precise time of that audio, then: Time Warp > Change > -20% > Save and you will effectively deepen a voice. Then open Time Warp again and select "Length" and restore the original time, be sure to also select "Similarity" and then hit save. You now have converted that voice to something different, it has a much lower pitch without goooiiinnngggg rreeeaaallll sssslllloooowwww.

It may be interesting, amusing and even fun, but it does not and will not sound like a real person speaking in a natural real person voice because it cannot sound like a real person speaking in a real person natural voice. In some cases it is possible to get almost, sorta, kinda close - but do not expect to make profound changes that are convincing enough to fool anyone. The electronic distortion - and that is precisely what is done in such a case - is apparent in all but the most subtle alterations. Smaller changes will not be so obvious and - if they are not too much altered from the original - may sound like a real person speaking in a real person's voice. But, that will not make a young girl sound like an adult man or any other such extreme and profound change.

I would urge anyone interested in this kind of thing to simply open up Goldwave, record a few words or sentence and then play around, have fun, experiment and make mistakes - YES - Make Mistakes because they teach us as much as our successes. Sometimes trial and error really and truly is a good way to learn.

[thudo]

Example: Highlight a bit of a track containing a few seconds of voice, note the precise time of that audio, then: Time Warp > Change > -20% > Save and you will effectively deepen a voice. Then open Time Warp again and select "Length" and restore the original time, be sure to also select "Similarity" and then hit save. You now have converted that voice to something different, it has a much lower pitch without goooiiinnngggg rreeeaaallll sssslllloooowwww.

Thanks Don! Thats precisely what I wish to achieve! I want more of that exactly..

[donrandall]

Thanks Don! Thats precisely what I wish to achieve! I want more of that exactly..

You can also play around with equalizer adjustments to attenuate certain frequencies and exaggerate others, but once again,that can produce some rather odd sounding voices rather than natural sounding real people voices, if not done with a light touch.

Female voices generally have most of their body, or content, or fundamentals between 200 and 1200 Hz, give or take a bit, with frequencies above that range being harmonics.

Experiment with increasing volume somewhere in the harmonics range to sharpen and add clarity, presence and life. Sorry, no one can tell you precisely where or how much, you will need to experiment with each particular voice.

If a female voice has too much bass or boom or has a dull, thick sound, try attenuating frequencies around 200 Hz, conversely, if you want more of that sound... you know what to do now, right?

Male voices generally have most of their body, or content, or fundamentals between 100 and 500 Hz, with frequencies above that range being harmonics.

Male voices with too much boom and bass and sound thick or dull, attenuate at around 100 Hz.

Add life or presence by finding a spot in the harmonics range where a little boost adds clarity, presence and life. Once again, every voice is different, so it is up to you to figure out just where and how much to boost.

Do you know what it meant by "sibilance"? That is the "S" sound that is too hot, too loud and tends to jump out at ya. Sometimes that can be tamed by equalization, it could be at 4000, or 5000, or somewhere higher. If you can figure it out, you can tame it with the equalizer. But understand, knocking down the volume in a frequency range also affects every other sound in that same frequency range, so you may discover that you can not benefit from either reducing sibilance or altering a voice characteristic. You may have to choose between the two. I have found that taming the sizzling hot "SSS" sound is done most effectively by zooming in and isolating only that sound, then cutting it by half - about -6dB and sometimes a bit more. Again, experimentation is probably going to be necessary.

[thudo]

Actually, can there be made pre-done FX Chains based on what you ask? Thats what I am looking for as you are getting into realms that most of us might require some assistance. Looking more for some scripts I can throw into latest Goldwave and experiment on my voice actor submissions for the multiple projects I am doing. Would be utterly massive help!

[donrandall]

Maybe someone else would know about that -- I really don't have any idea.

I do know that for any effect you use to alter a voice, you can save the settings as a preset. Once you have everything adjusted as you want it for a particular effect, you simply type a name in the white field, the Plus Sing will the light up, click on the Plus Sign and the preset is saved.

[thudo]

Oh I know I can save preset FX Chains.. I was just curious if someone had a set already done. I'll create my own of course.

[donrandall]

Here's some websites you might find interesting:

http://acoustica.com/kb/How-can-I-chang ... _1119.html

http://emusician.com/dsp/antares_voice_thing_0221/

[thudo]

Thanks Don for that.. Antares Autotune looks more interesting as its a VSTi then Voice Thing which does real-time voice warping (which I don't want because I have voice actors doing all my recordings with their raw voice so as a post-processor I want to be able to batch-change tons of voices to suit a single unit type quickly). So I take, say, 60 wav files for a single in-game unit, then use either Goldwave, SoundForge, SoundBooth, etc and mass-batch all the various wav files of the same unit so they sound different. Its critical I can do this was all my wav files and NOT voice acting with a mic myself: I'm not ready to do this myself although I can easily go out and by a good recording USB Mic + DIY Soundbooth to put the high-end USB Mic into the sound-proof portable box. However, I'm busy enough with coordinating and post-processing.

Thanks Don.. magnificent finds!

[donrandall]

I hope it all works well for you!

[thudo]

err. thread necro?

Btw, still figuring it out.. Due to the amount of work I need done, another post-processor would help me get through this huge queue.

[thudo]

More spam?

Anyway, using multiple programs as well as GoldWave to do what I need to do but generally working in Warhammer 40,000 universe is quite forgiving as most voices are behind a mechanical vox so it can help mask a voice well. So far going well.

[thudo]

Indeed except that is beyond my scope unless I can get VST addons to work with Goldwave (which I believe does not work) or competitors which do (Soundforge). Pitch is part of what I do to change a voice. Adding vox and FX via VSTs is another part. Warhammer 40,000 is quite forgiving when being able to apply masks behind even damaged voices so they sound hi-tech.

[Torsten]

In that case a natural sound is vital, but there is no need to support changes greater than about 20 or 30% either way, as anything more than that stands practically no chance of sounding natural anyway.

[jpnullnvoid]

first, a disclaimer: the process i am about to describe is extremely tedious and the results are dependent upon a lot of factors. I DO NOT GUARANTEE SATISFACTION! that being said, i've had some relative success altering formants in speech using the Noise Reduction tool.

1.) start with two speech files, one will be the source and the other will be the target. note: these files should be roughly the same volume.

2.) select a section of the target file containing the formant (vowel sound) that you want to alter. then, find and select a sample of the same vowel in the source file.

3.) copy the source vowel to the clipboard and re-focus on the target file.

4.) open the Noise Reduction tool, select Clipboard as your noise source and select the option listen to the noise profile.

5.) here's the tricky part - you will need to play around with the scale and FFT parameters to get a sound that you MIGHT find acceptable. i usually find lowering the scale to somewhere below 50 works best.

6.) volume adjustments and pop-click removal may be necessary after processing with the NR tool.

this method is by no means perfect and can be VERY time-consuming, but i've gotten some interesting results morphing my wife's voice with my own (no, i don't have a saved example at the moment but i'll try to get one uploaded to my SoundCloud and share it with you). my point in posting this is simply to demonstrate that GoldWave CAN do things beyond the basic audio editor stuff. you just have to be willing to get creative and experiment.