Event, Meeting > Whose voice is it anyway? [London]

Title: Whose voice is it anyway? [London]
Location: Royal College of Pathologists, 2 Carlton House Terrace, London, SW1Y 5AF
Description: Lecture by Jeff Bloom, Synchro Arts
Start Time: 18:30 for 19:00
Date: Tuesday 13th March 2012

Abstract:

Editing audio to fix timing and tuning problems has now become so commonplace that listeners would be hard-pressed to know when the timing, pitch or other characteristics of a recorded actor or singer have been manipulated to be more accurate or to simply sound better.

However, even with sophisticated software tools, in many situations the editing work required to achieve such polished precision can still be tedious and time consuming, and require considerable skill.

In this talk new processing techniques will be demonstrated which offer automated and precise solutions to certain common situations. These techniques involve automatically extracting, from an accurate ‘guide’ voice or instrument recording, selected characteristics such as timing, pitch, vibrato and loudness, and imposing these features on other less accurate recordings of similar performances.

This approach has many applications in consumer and professional audio processing products, including the following…

For professional applications:
1)     Double and triple (or more) tracks can be made quickly to match, with adjustable precision, an accurate lead vocal.
2)     Alternative performance characteristics can be transferred to a lead vocal.
3)     Prosodic features (including timing, inflection and stress) of recorded dialogue can be transformed to have different but natural sounding features transferred from another recording.

For consumers:
4)     In websites or mobile applications, recordings of amateur singers can be automatically transformed to have the characteristics of a professional vocalist.
5)     A language student’s recording of his or her attempt at mimicking a teacher’s recorded sentence can be modified to make the student’s timing and pitch sound like the teacher’s and provide constructive feedback.

Jeff Bloom – who in 1984 invented the first audio time-alignment algorithms upon which these new techniques are based – will also chart the history of automatic time alignment in dialogue replacement and music applications.