Pitch Shift/Time Stretch to new tempo/key

Sonar allows to transpose and stretch audio in many ways. Recently I tryed to fit some vocal phrase into a project. Original vocal was taken from sampling CD. It was in F major and 140 bpm. I created remix project in the key of Ab major and with tempo 135 bpm. Therefore I needed to transpose audio by minor 3rd up and stretch it from 140 to 135 bpm. I also wanted the formant of the vocal not to be shifted.

Original vocal phrase:
http://www.box.net/public/hxsajtmrfp

First I select the clip, open clip properties and check "Stretch to project tempo".


Then I double-click the clip and set Transients Detection at 100% and Pitch at +3.


The result:
http://www.box.net/public/jl963gl3db

There are echoes, and the formant is shifted, making munchkin effect.

Next method I tryed was Transpose Audio from Process menu.


There is formant shift, but it is expressed in terms of 10th of octave! What should I enter in this field if I want the formant not to be shifted? Vocal is transposed up by minor 3rd; there are four minor thirds in octave; so I enter -2.5 and choose Radius Vocal algorhythm.
http://www.box.net/public/62ft04osul

The quality of pitch-shifting is good, but what's with the formant? It's wrong. You can't simply keep formant at original pitch. You have to shift it in opposite direction and you have to deal with decimal fractures of octave, which are not musical intervals. There is no time-stretch option, so to fit the phrase into new tempo I slip-stretch the clip. Slip-stretching expresses length in percents. The question is: by how many percent should I stretch the clip to change its tempo from 140 to 135 bpm? Calculation shows 104%.


Quality of stretching in this case is good,
http://www.box.net/public/fsliced2mo
but if you'll try bigger percentage, you'll hear audible artefacts.

Next method I tryed was V-Vocal. It handles formant right way - by semitones, and transposing is easy.


There is also time-stretching, but you don't know exact amount of stretching in order to get from one tempo to another.


Sound quality of transposition in V-Vocal is good in most cases.
http://www.box.net/public/sx2lls4mb9
Sometimes, if audio level is low, V-Vocal produses artefacts. It does not recognize low level audio signal.

My last attempt was using Cakewalk Time Pitch Stretch 2 plugin.


Again, tempo change expressed in terms of percentage and formant shift in terms of decimal fractures. Why? Musicians don't think percentage and 10th of octave, they think BPM and semitones.

Here is the result:
http://www.box.net/public/gi108imxkb

(I intentionally omit Audio Snap method. My experience with Audio Snap on vocals was completely unsatisfactory, especially with Radius Vocal type of stretching. I always got echoes and artefacts.)


So, what's the point?
In Sonar we have 5 different methods of pitch(formant)-shifting and time stretching.

1. Stretch to Project Tempo. Pros: fits clip into new tempo automaticaly, allows pitch shifting. Cons: no formant-shifting, unacceptable audio quality.

2. Process > Transpose Audio. Pros: easy to use, allows formant shift. Cons: no time-stretching, confusing formant-shift input, Radius produces strong artefacts if transposes to wide intervals.

3. Slip Stretch. Pros: easy, intuitive way of stretching. Cons: no pitch-shifting, percentage input instead of tempos, Radius algorhythm produces audible artefacts if stretches more than 120%.

4. V-Vocal. Pros: proper formant handling, good audio quality. Cons: doesn't recognize low-level waveforms, no way to figure out exact stretching amount in order to fit to new tempo.

5. Cakewalk Time Pitch Stretch 2. Pros: best sound quality of five reviewed. Cons: percentage instead of tempo, confusing decimal fracture formant-shift input.

My question is: why five? Cakewalk, please give us simple and intuitive tool for simultaneous transposing and stretching of audio with ability to keep formant. With simple interface which might be looking like this

I agree about pictures, that was big mistake. I just wanted to keep all the examples in one place. Tonight I'll fix it.

Formant shift is important part of audio pitch-shifting. Formants are certain frequencies which give the voice its unique character. Male vocal's formants differ from female ones. If you transpose vocals, all the formants follow transposition. As a result you have either Munchkin or Dart Vader depending on which direction you transpose. To avoid this sindrom, pitch-shifting tool must allow you not to shift formants, so the character of the voice remains unchanged.

ORIGINAL: Vovchik

I just fixed picture display, I hope the message is more readable now.

Thanks, it's much easier to read and watch now!

Formant shift is important part of audio pitch-shifting. Formants are certain frequencies which give the voice its unique character

Thanks for the reply, I do know the basics I guess.. below you gave some more details about how to adjust the formant shifting and that what i was actually had in mind, I just use formant shifting by ear more or less.. I don't think it's that bad ;)

I see your point about the values there (Decimal) now -

There is formant shift, but it is expressed in terms of 10th of octave! What should I enter in this field if I want the formant not to be shifted?

The quality of pitch-shifting is good, but what's with the formant? It's wrong. You can't simply keep formant at original pitch. You have to shift it in opposite direction and you have to deal with decimal fractures of octave, which are not musical intervals

in V-Vocal -

There is also time-stretching, but you don't know exact amount of stretching in order to get from one tempo to another

yup, anyone has the answers for these?