What is Pitch Correction: an easy explanation Some people think that anyone who can produce reasonable vocal sound can be made to sound like a good singer with modern technology. That is not really true, though a lot can be done nowadays to improve for instance a recorded vocal by performing pitch correction. When a vocal is good in terms of sound and feeling and there are only a few off-key notes, it may be worth while to perform pitch correction, if only to save time and money by not having to do the recording again, even if you could. Here we try to explain wat pitch correction does in not too techical terms, though a few technical words are inevitable. First of all, pitch is not equivalent to frequency, but closely related to it; it is a perceptual quality that is primarily determined by the fundamental frequency of a natural tone. Any tone produced by a voice or instrument contains harmonically related frequencies which are a multiple of the fundamental frequency. For instance, an ''A'' at 220 Hz (i.e. 220 vibrations per second) contains also frequencies of 440 Hz, 660, 880 Hz etc; this is just how vocal chords, strings etc vibrate naturally. In this example 220 Hz is the fundamental frequency and this we perceive as the pitch. However, the perceived pitch is also influenced a little by duration of a note and loudness; a very loud note tends to sound a bit higher in pitch. Then there is also some noise (e.g. breath noise); the noise does not have a pitch. When a singer sings ''A" off-key, it might for instance consist of 215 Hz, 430, 645, 860 Hz etc. To restore the correct pitch, we cannot just shift all frequencies up by 5 Hz, because we woud get a wrong harmonic sequence, which sounds extremely bad, and with no clear pitch at all. So how does it work? The simplest way is to speed up the sound, by playing it faster than it is recorded. This will make it sound a bit faster, but also higher in pitch. A big disadvantage is that a sound that is played back faster, also becomes shorter in time. In order to compensate for the change in duration, we can time stretch the sound, by cutting it into a lot of very short parts (like a few hundred per second) and then adding some of those same additional parts to make the sound longer again (or shorter by overlapping the parts). This way we can change pitch and preserve the duration. A second disadvantage of changing the speed is that the pitch is changed, but also the timbre or colour of the sound. An adult voice will start to sound like a child's voice if pitched up this way. Why is this so? It has to do with so-called formants. A vocal sound is produced by a singer's vocal chords, but is colored also by the throat, mouth etc. These colorations are the formants. In technical terms, this is called the source-filter model. Now when a singer raises his/her pitch, the vocal chords (source) produce higher frequencies, but the ''coloration'' (filter) remains approximately the same, as the throat and mouth do not change. In contrast, when we just speed up a sound, both the pitch and the coloration are changed and so it sounds like a voice of someone with a smaller throat and mouth, like a child. So, a second improvement is trying to preserve the formants. This can be done by first measuring the coloration (formants), and taking it out by applying an inverse coloration. Then, the ''uncolored'' neutral pitched sound is pitch corrected, and the original coloration is applied again to this. Now we have a good approximation of what the singer would have produced by singing a higher pitch. There are a lot of technical diffulties and details to be taken into account in applying the above principles, but all modern pitch correction techniques have incorpated such methods, one way or the other. Pitch correction is mostly used in studios on vocal tracks, and sometimes even live on stage. Until now, correction of vocal pitch of a voice inside an already mixed recording was not possible, because a pitch correction would pitch shift the whole music. CSP audio has a developed a new technique that can pitch correct a voice inside a mix, see www.pitchcorrection.com.
For those who are technically inclined here are some technical links about pitch correction: |
