Corpus-Based Concatenative Synthesis

Schwarz, Diemo

doi:10.1109/msp.2007.323274

Cited by 85 publications

(76 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Corpus-based concatenative synthesis has been implement in the software CataRT, with its associated signal processing package FTM & Co., by Diemo Schwarz and the Sound Music Movement Interaction team at IRCAM (Schwarz 2007). CataRT allows for the analysis and segmentation of a database of samples recorded live or in deferred time, the corpus, and resynthesis through to a variety of control paradigms.…”

Section: Concatenative Synthesismentioning

confidence: 99%

Mapping the Klangdom Live: Cartographies for Piano with Two Performers and Electronics

Einbond

2017

Computer Music Journal

View full text Add to dashboard Cite

This is the published version of the paper.This version of the publication may differ from the final published version. Abstract: The use of high-density loudspeaker arrays (HDLAs) has recently experienced rapid growth in a wide variety of technical and aesthetic approaches. Still less explored, however, are applications to interactive music with live acoustic instruments. How can immersive spatialization accompany an instrument already with its own rich spatial diffusion pattern, like the grand piano, in the context of a score-based concert work? Potential models include treating the spatialized electronic sound in analogy to the diffusion pattern of the instrument, with spatial dimensions parametrized as functions of timbral features. Another approach is to map the concert hall as a three-dimensional projection of the instrument's internal physical layout, a kind of virtual sonic microscope. Or, the diffusion of electronic spatial sound can be treated as an independent polyphonic element, complementary to but not dependent upon the instrument's own spatial characteristics. Cartographies (2014), for piano with two performers and electronics, explores each of these models individually and in combination, as well as their technical implementation with the Meyer Sound Matrix3 system of the Südwestrundfunk Experimentalstudio in Freiburg, Germany, and the 43.4-channel Klangdom of the Institut für Musik und Akustik at the Zentrum für Kunst und Media in Karlsruhe, Germany. The process of composing, producing, and performing the work raises intriguing questions, and invaluable hints, for the composition and performance of live interactive works with HDLAs in the future. Permanent repository link

show abstract

Section: Concatenative Synthesismentioning

confidence: 99%

Mapping the Klangdom Live: Cartographies for Piano with Two Performers and Electronics

Einbond

2017

Computer Music Journal

View full text Add to dashboard Cite

show abstract

“…Pruning spurious units improves the TTS output [37,[47][48][49][50][51] while pruning redundant units reduces database size thus enabling portability [52][53][54] and real-time concatenative synthesis [2,55,56]. In this work, we focus on removing spurious units and not redundant units.…”

Section: Data Pruning Using Confidence Measuresmentioning

confidence: 99%

Developing a unit selection voice given audio without corresponding text

Godambe

Rallabandi

Gangashetty

et al. 2016

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

Today, a large amount of audio data is available on the web in the form of audiobooks, podcasts, video lectures, video blogs, news bulletins, etc. In addition, we can effortlessly record and store audio data such as a read, lecture, or impromptu speech on handheld devices. These data are rich in prosody and provide a plethora of voices to choose from, and their availability can significantly reduce the overhead of data preparation and help rapid building of synthetic voices. But, a few problems are associated with readily using this data such as (1) these audio files are generally long, and audio-transcription alignment is memory intensive; (2) precise corresponding transcriptions are unavailable, (3) many times, no transcriptions are available at all; (4) the audio may contain dis-fluencies and non-speech noises, since they are not specifically recorded for building synthetic voices; and (5) if we obtain automatic transcripts, they will not be error free. Earlier works on long audio alignment addressing the first and second issue generally preferred reasonable transcripts and mainly focused on (1) less manual intervention, (2) mispronunciation detection, and (3) segmentation error recovery. In this work, we use a large vocabulary public domain automatic speech recognition (ASR) system to obtain transcripts, followed by confidence measure-based data pruning which together address the five issues with the found data and also ensure the above three points. For proof of concept, we build voices in the English language using an audiobook (read speech) in a female voice from LibriVox and a lecture (spontaneous speech) in a male voice from Coursera, using both reference and hypotheses transcriptions, and evaluate them in terms of intelligibility and naturalness with the help of a perceptual listening test on the Blizzard 2013 corpus.

show abstract

“…The user can also rely on two algorithms to automatically dispose the objects. The first one, inspired by cataRT software [18], calculates the objects' positions and colors according to descriptors chosen by the user. The second calculates the positions depending on a sample of objects selected by the user.…”

Section: How Could Idda and Big Data Help Supporting Our Figural Comentioning

confidence: 99%