To its credit, the Lexicon 2400 was used by Stanley Kubrick in the making of Full Metal Jacket. Mr. Kubrick was compelled write an unsolicited thank-you note to Lexicon detailing one scene where it was used creatively; where a rocket explodes into a building in slow motion. That credit to the Lexicon Time Compressor/Expander appears at movie's end (≈1:56:15). Other credits include the animations An American Tail and The Simpsons.
I designed all the algorithms for this unit including a pitch detector and a splicing algorithm. I designed the SPLICE board and Confidence board that together constituted the digital signal processing engine. The splicing algorithm had an entire board dedicated to its execution, as did the pitch detection algorithm: the Confidence board. I also designed the crystal oscillator that gave the unit timing accuracy to three decimal places (pushing TTL technology). [Dattorro, Simple Techniques Extend Counter Range, Design Ideas, EDN magazine p.307, July 25 1985] The unit also housed a HOST board for user interface, and a machine-control board to interface with tape machines (designed by Manny Guerrero). Chuck Bagnaschi designed the beautiful front panel that made the unit very easy to use.
The splice engine comprised two TMS32010 DSP chips. The SPLICE board queried the confidence board mathematical coprocessor. So there were a total of four DSP chips in this design. The SPLICE board also performed digital filtering to compensate for aberrations caused by speeding or slowing tape machines of the type used by the television broadcast industry. The digital filters employed error feedback correction as discussed in my paper http://ccrma.stanford.edu/~dattorro/HiFi.pdf
The Lexicon 2400 has been reverse engineered many times; most notably by EvenTide Inc. Introduced in 1986, it remained in production through 1996. It also remains the standard for time compression in the United States and France. DILATATION ET TRANSPOSITION SOUS CONTRAINTES PERCEPTIVES DES SIGNAUX AUDIO by Gregory Pallone.
AUDIO QUALITY USER SETTING:
SOLO VOICE/INSTRUMENT has an average system propagation delay of 28ms (0.8 frames NTSC, 0.7 frames PAL/SECAM). It offers fine temporal resolution, and is most useful for solo material (single voice or instrument) such as solo saxophone, solo violin, speech, or highly transient material such as solo drums. It is especially useful for picking out speech in the presence of strong background noise or music ("voice-over" commercials, for example) because the DSP "locks" onto the speech frequency band and preferentially splices there. In general, solo material sounds better using SOLO VOICE/INSTRUMENT because the onsets in the signal envelopes are more accurately detected. Certain types of fast moving, polyphonic music having many rapid attack transients, however, may also sound better using this analysis.
POLYPHONIC has an average system propagation delay of 59ms (1.8 frames NTSC, 1.5 frames PAL/SECAM). It offers fine frequency resolution, and is most useful for highly complex material containing mainly musically correlated sound sources. Examples would include symphony orchestra, rock music, etc. It should be used when the input signal contains significant low frequency energy, such as in solo bass guitar. POLYPHONIC will splice speech and solo instruments very well but, for the most demanding solo applications, SOLO VOICE/INSTRUMENT should be used.