0
\$\begingroup\$

In music, vibrato describes the warble effect often added in by singers as they sing. I was trying to come up with a circuit to remove it from an audio signal, almost like a "vocal stabilizer" leaving only the pure vocals and I'm a bit stuck. My first thought was that there are a few characteristics of vibrato that differentiate it from the rest of the signal. Most of my knowledge of audio circuitry is self-taught, so please excuse my lack of proper terminology. Vibrato tends to be:

  • One relatively-regular low frequency (~2-20Hz) overlaid upon the existing "note frequency" (the 80Hz-15kHz normal range of human voice determining note in the music)

  • Oscillation of both amplitude and frequency, sharing the same interval. I.e. in vibrato both the volume and note go up and down simultaneously, at that same 2-20Hz

  • Infrequent, only appearing on certain parts of the signal (by time)

It would seem based on these three characteristics that vibrato is definitely differentiable, it's just difficult to extract without removing other detail. Any ideas?

\$\endgroup\$
  • 2
    \$\begingroup\$ This would be extremely difficult to do with an analog circuit, but is possible with DSP. There are lots of pitch correction tools used in audio production that can add and remove vibrato and shift the pitch of recorded material. The pitch can be corrected real-time, though I'm not sure if you could remove vibrato real time effectively without causing artifacts. It's certainly possible after recording though. Check out Melodyne, Auto-tune, and V-Vocal for examples. \$\endgroup\$ – John D Jan 9 '16 at 17:56
1
\$\begingroup\$

One relatively-regular low frequency (~2-20Hz) overlaid upon the existing "note frequency" (the 80Hz-15kHz normal range of human voice determining note in the music)

This is not entirely correct. Vibrato is not a 20Hz component summed with the vocals, it's actually frequency modulation: the LF component is modulating the vocal range.

To remove vibrato, you could try use autotune with a short response time (to stabilize frequency) and compression (to stabilize amplitude). Careful tuning of both these effects would likely do a decent job of removing vibrato. However, note that this is not something that can be done with a simple analog circuit. Pitch-shifting of complex signals such as vocals (let alone autotune) is decidedly non-trivial to implement in the analog domain.

\$\endgroup\$
  • \$\begingroup\$ Note that "autotune" is a function that is available in many sound processing applications (hardware and/or software) for live and recorded music, so it's relatively easy to experiment with. However, note that some singers use a very "deep" vibrato -- one that spans many semitones -- which would probably give an autotuner fits. \$\endgroup\$ – Dave Tweed Jan 9 '16 at 18:05
  • \$\begingroup\$ @DaveTweed Yup, that's a good point. I would guess that a regular autotune algorithm could be adapted to handle those cases, but it would certainly require a lot of tricky programming to differentiate between vibrato pitch changes and "regular" pitch changes. \$\endgroup\$ – uint128_t Jan 9 '16 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.