Skip to main content ↑ ↓
Topic: Audio Pitch Shifting (Read 3145 times) previous topic - next topic

Audio Pitch Shifting

I have recorded an instrumental track (treble recorder) against a pre-recorded Noteworthy accompaniment. But the recorder was not warmed up enough when it was played and on the recording it's a bit flat. Does anyone know of a free tool which can sharpen-up the audio (.WAV) file a little bit without changing the playing speed?

Thanks. David.

Re: Audio Pitch Shifting

Reply #1 –
Pitch Shifting is generally a standard feature in audio editors. You might try searching for audio editors and see if any meet your requirements (pitch shift, free, etc...).

Re: Audio Pitch Shifting

Reply #2 –
The free "Audacity" (audacity.sourceforge.net) can do that.

Re: Audio Pitch Shifting

Reply #3 –
If your sound is a recorder (resonance in a tract) then you can use Phonotron 1 (http://www.phonotron.com) to do a natural pitch shift without changing the speed. Natural pitch shift (formant preserving) will make the recorder sound like how it would do if you played it at the pitch shifted note, whereas normal pitchshifting modifies the 'colour' of the tone too. Although you probably actually need normal shifting in your case, at least you can play with the actual melody too in Phonotron which you cant do with a normal shifter on those kinds of sounds.
 

Re: Audio Pitch Shifting

Reply #4 –
The natural pitch shifting deserves a few more remarks.

The simplest pitch shift also makes music go faster. Also, unless the change is very small, it will sound quite wrong. The analogy is to play a 33 RRM vinyl record at 45 RPM.

A somewhat improved algorithm detects the sudden changes in pitch that mark when a note is changed. It preserves the time points of these changes, then analyzes the pitches in between time points. These pitches are changed, but the time duration is preserved. So the music plays at different pitch, but still maintains the same beat. However, unless the pitch change is small, it will sound quite wrong - just as in the first case.

The "formant" method is based on physical modeling of the particular instrument. Standard instruments have known physical models. It is also possible to use this method for the human vocal tract, with different models for different phonemes - except that good singers change the vocal tract depending on pitch range and nuance, so no single set of models works.

When pitch is changed by the formant method, the relative amplitudes of fundamental and overtones are adjusted, not just their pitches. This produces a much more natural sounding effect.

So, if the music is a particular solo instrument, the best way to shift pitch would be to use software based on formants. But if the music has many unrelated instruments, there would be no reliable way to apply the different formant models.