Dolby Laboratories 585 6.2 Development of Time Scaling

Model 585 Time Scaling Processor Principles of Operation

6-2

6.2 Development of Time Scaling

Time scaling of audio was first achieved in the 1950s with the invention by

Fairbanks, Everitt, and Jaeger1 of a modified tape recorder. Their modified machine

discarded and repeated sections of audio signal by using a rotating playback head, and

became known as the sampling or splicing technique.

Based on a time-domain representation of the signal, time-domain techniques are

efficient and have been attractive for use in real-time implementations. The Fairbanks

splicing technique was a very low-fidelity solution, producing a choppy product that

sometimes obscured the content.

Lee2 translated the splicing technique into a digital system in 1972, replacing the

magnetic tape with a circular memory buffe r and using read/write memory pointers in

place of the rotating head. Further advances were made on this solu tion throughout

the ’70s, ’80s, and ’90s. These advances generally applied signal analysis to the

splicing technique, making use of increased computer processing power during that

time period. The solutions developed tended to focus on either speech or musical

instruments, but not combinations of both.

The more recent solutions are significant improvem ents on the original splicing

technique, but all time-domain based processors exhibit a characteristic drawback that

at minimum makes the listener aware that the audio has been altered. Characteristic

artifacts of these methods often include audible echoes, stutters and an overall loss of

definition in the audio.

In 1967, an alternative to time-domain techniques was introduced by Schroeder3. This

method used either a Fourier transform or a filter bank to produce a frequency

representation of the signal. A direct manipulation of the frequency components of

the signal was followed by a transformation back into th e time domain to achieve

time scaling.

The technique introduced by Schroeder was developed for speech. Later

modifications applied this approach to music, and variations were produced that could

handle specific signal types, either music- or speech-oriented, but still not in

combination.

1 Fairbanks, G., W. Everitt, and R. P. Jaeger. “Method for Time or Frequency Compression-Expansion of Speech.”

Transactions of the Institute of Radio Engineers, Professional Group on Audio AU-2 (1954): 7–12.

2 Lee, Francis F. “Time Compression and Expansion of Speech by the Sampl i ng Me th od .” Journal of the Audio

Engineering Society 20, no. 9 (1972): 738–742.

3 Schroeder, M., J. Flanagan and E. Lundry. “Band-Width Compression of Speech by Analytic-Signal Rooting.”

Proceedings of the IEEE 55 (1967): 396–401.