User:Memory 001/test

DAISY Project is the code name for the VOCALOID development project launched by Yamaha in March 2000. The name was coined in 1961 as a result of an open experiment at Bell Labs in <! -- IBM 704 --> the world's first computer to sing<! -- sung and quoted in the 1969 film 2001: A Space Odyssey -->song "Daisy Bell". <! -- ref name=togetter11/ -->The official name of the <! -->The official name of the product was decided to be "VOCALOID" and was officially announced in February 2003.

Background
April 2000 A collaboration with the Pompeu Fabra University Music Technology Group (MTG) of Barcelona was<! -- As part of the project, the signal processing part of VOCALOID was developed In May 2002, contacts were initiated with Crypton Future Media in Sapporo, Zero-G Limited in England in the autumn of the same year, and one other company, and later (with at least the two aforementioned companies Later (with at least two of the aforementioned companies), a licensing agreement was reached for the production of singing voice libraries and software sales. A press release about the development on February 26, 2003, and in March of the same year, after prototype exhibits and presentations at the Musikmesse and AES Convention, in January 2004, at the NAMM Show The first VOCALOID products, [[LEON/LOLA|Leon and Lola, were announced by Zero-G and released in Japan on March 3 of the same year.

Music Technology Group
[[Image:Reactable Multitouch.jpg|thumb|120px|Reactable]

The Pompeu Fabra University Music Technology Group (MTG), with which Yamaha collaborated, is a research group on sound and music computing founded in 1994, and currently has about 40 researchers. The virtual modular synthesizer with a real-world interface Reactable one of the results of MTG's research and development. Other known activities include Freesound Project (Freesound.org) and BMAT, a music-related IT company.

Xavier Serra
[[Image:Xavier Serra 1, Music Hack Day Barcelona 2012.jpg|thumb|90px|Xavier Serra]

MTG founder and director, Xavier Serra, was a member of the Stanford UniversityCCRMA in the 1980s, and has been a member of the He has been working with Julius O. Smith, a well-known physical modeling and synthesis, on a pitch-synthesis method for phase vocoders similar to the MQ method. In 1987, he developed an analysis/synthesis method PARSHL using a phase-vocoder pitch tracking extension similar to the MQ method. Also in 1989, McAuley and Quatieri proposed a sinusoidal-based speech analysis/synthesis method Sinusoidal modeling is an extension of the speech synthesis acoustic model to include inharmonic noise components, which are proven in speech synthesis. In this paper, we propose an inharmonic music analysis/synthesis method Spectral modeling synthesis (SMS) that adds noise components, which have been proven in speech synthesis, to acoustic models. This SMS method is also used as one of the basic technologies in the joint research on VOCALOID that started in April 2000.

Results of joint research
According to, the signal processing methods developed in the MTG and Yamaha collaboration are described in three papers from 2001-2003 , , and. In this study, we used frame-based frequency-domain techniques (i.e., frequency-domain processing of frame-by-frame audio fragments, such as diphone) to process <! -- In this paper, a system that synthesizes singing voices by transposition/time stretching/concatenation/waveform connection synthesis/waveform connection synthesis/waveform connection synthesis] is presented.

The speech model in this study is based on a "harmonic+residual" representation using the SMS method, one of the spectral models, and is a novel extension of the source filter model, one of the quasi-physical models. developed <! -- It is based on the "excitation plus resonance" representation of the newly developed Excitation plus Resonances (EpR) speech model, which is an extension of the source filter model, one of the quasi-physical models. The difference between the model and the original waveform is the difference in spectral shape <! -- (differential spectral shape)--> during analysis and added during re-synthesis to suppress sound quality changes <! -- ref> (PDF).

As a basis for singing voice synthesis, phase-locked vocoder based on Sample deformation method using the frame-based spectral analysis/synthesis method Spectral peak processing (SPP) &mdash;&mdash; time scaling, pitch transformation with nonlinear scaling of the spectrum, phase correction, and peak intensity adjustment of the spectral envelope (equalization) for tone adjustment &mdash;&mdash; was developed. For the feature connections, we inserted transition frames between the feature frames and used the above sample transformation technique to create phase connections<! -- (Phase concatenation)--> and spectral shape connection<! -- (Spectral shape concatenation)--> (the so-called spectral envelope interpolation ) using the above sample deformation method was developed.

The technology actually used in the commercial version of VOCALOID is outlined in, for example,.

Related items

 * VOCALOID
 * Vocal Synthesizer
 * Waveform connection synthesis
 * Spectral modeling synthesis (SMS)
 * Phase vocoder