DILATATION ET TRANSPOSITION SOUS CONTRAINTES PERCEPTIVES DES SIGNAUX AUDIO : APPLICATION AU TRANSFERT CINEMA-VIDEO

Abstract : Coexistence of different formats for cinema (24 frames/s) and video (25 frames/s) involves speeding up or slowing down the soundtrack when converting from one format to another. This causes a temporal modification of the sound signal, and therefore a spectral modification with a change in timbre. Audiovisual post-production studios have to compensate this effect by an appropriate sound transformation. The aim of this work is to propose to the audiovisual industry a system which allows the counteraction of timbre modification caused by a change in the playback rate. This system consists of a processing algorithm and a machine on which it is implemented. The algorithm is designed to respect sound quality and multichannel compatibility constraints. The machine, named HARMO, is designed for this purpose by the company GENESIS. It is based on digital signal processors and has to respect real-time constraints. The commercial aspect of the project is linked to economic and timing constraints. A state of the art based on a quasi-exhaustive bibliography leads to an original classification of existing time-stretching and pitch-shifting methods. Well-known time-domain and frequency-domain methods are studied, and time-frequency methods are introduced. This classification allows the creation of several innovative methods: . two time-frequency methods using an analysis technique adapted to the human ear, . two coupled methods using advantages of both time- and frequency-domain methods, . a method which proposes an improvement of time-domain methods. Algorithms are evaluated using a bank of test sounds specially designed to highlight characteristic artifacts. The time-domain approach is selected and optimized thanks to criteria based on normalized autocorrelation and detection of transients. This algorithm is integrated into a software designed for multichannel real-time running, and implemented on the HARMO hardware.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00003363
Contributor : Grégory Pallone <>
Submitted on : Tuesday, September 16, 2003 - 8:58:33 PM
Last modification on : Monday, March 4, 2019 - 2:04:03 PM
Long-term archiving on : Monday, September 20, 2010 - 11:36:24 AM

Files


Identifiers

  • HAL Id : tel-00003363, version 2

Citation

Grégory Pallone. DILATATION ET TRANSPOSITION SOUS CONTRAINTES PERCEPTIVES DES SIGNAUX AUDIO : APPLICATION AU TRANSFERT CINEMA-VIDEO. Modélisation et simulation. Université de la Méditerranée - Aix-Marseille II, 2003. Français. ⟨tel-00003363v2⟩

Share

Metrics

Record views

10

Files downloads

464