Check out the beta version of the new IRCAM Forum ⇢

TribuneView more

Making an orchestra speak


Speaking is a piece for orchestra and live electronics that employs various offline and real-time technologies for creating a hybrid space between orchestral and speech sounds.

 “Speakings” is a piece for live electronics and large orchestra composed by Jonathan Harvey in 2008 and in collaboration with computer musicians Gilbert Nouno and Arshia Cont. The world premiere occured in August 2008 in Royal Albert Hall of London by BBC Scotish Orchestra. The composition was awarded the Prince Pierre of Monaco Prize in Musical Composition in 2009. A recording by BBC Scottish Symphony is available on Outhere Records. And a paper published in the 2009 Sound & Music Computing conference on the making of the piece was awarded the best paper.

The realization of “Speakings” was undertaken in two stages: The first, through Computer Aided-Composition and as an aid in preparing score materials; and second, real-time effects and performance realizations. On top of these were added various effects and controls not discussed here. We discuss the two phases below and share Demo Patches in the attachment to reproduce each item:

Voice to Music Score Transcription

In the first phase using OpenMusic and its OM-SuperVP library, selected voice soundfiles were transcribed to music scores. The OpenMusic patch is shown at the right. It uses SuperVP capabilities to analyze voice and capture partials, and uses RepMus Libraries to transcribe them to notes and quantize them rhythmically. This OpenMusic patch is available in the Attachment.

In addition, the Orchidée system suggested several orchestration schemes over selected patterns. These two steps provided enough material for the composer to create an orchestral score.

Stamping Voice Envelopes on Live Orchestra

While transcription and automatic orchestration provide orchestral patterns close to speech structures, they are far from rich spectral morphologies observed in speech. To reach these morphologies, a real-time algorithm based on FTM technologies was developped to enrich the spectral content of real-time orchestra using provided speech. The patch demonstrating this live process is available in the Attachment.


Speakings by J. Harvey, Performance Scheme

Live Performance

To assure synchronicity between live electronics and the orchestra during live performance, electronic queues are tracked via a keyboard player placed in the orchestra using score following technologies. Through Antescofo, the tempo of the performance is tracked and driven into SuperVP for Max and Gabor technologies in FTM for best synchronization of live treatements and the orchestra. In addition to this, a computer musician controls live spatializations and movements as well as electronic levels through a controller.

During live performance, computer music performers mix and control spatial parameters using multi-touch interfaces communicing with Max patches and live processings synchronize with the orchestra following the tempo and position of the orchestra with the electronics.

For more details, refer to the article by the composer and involved musicians and researchers published in SMC 2009 proceedings available online.


5 Responses to “Making an orchestra speak”

  1. external libraries of open music.
    Where can I find this?
    Geert De Praetere

  2. Profile photo of Arshia Cont says:

    Hi Stad,
    Just click on the OM-SuperVP library link in the article or go here:

  3. Profile photo of trazomtg trazomtg says:


    i don’t understand really what you say.
    is it possible with OpenMusic to get sounds from a music instrument and transform these sounds in notes on paper partition?

    Please help!!!!!

    • Profile photo of alberto alberto says:

      That’s what Orchids software does. It is not exactly made in OpenMusic, but you can export data from Orchids. The Orchidee software mentioned in the paper has evolved to Orchids (so it is not anymore an OpenMusic object, but an external program)

      • Profile photo of Arshia Cont says:

        To complete: Orchids (or the prior Orchidé) was used in Speakings as a separate/external software and in addition to OpenMusic. We used OpenMusic to do analysis of speech audio and to get a music score representing raw harmonics. This can be reproduced by the OM Patch that is available for download on this Tribune. Orchids is used separately to propose possible orchestrations of the speech audio.
        You can also read the linked SMC Paper on this post for more details.

Leave a Reply

You must be logged in to post a comment.