EPSRC Reference: |
GR/J53355/01 |
Title: |
ADVANCED SPEECH CODING SYSTEMS |
Principal Investigator: |
Xydeas, Professor C |
Other Investigators: |
|
Researcher Co-Investigators: |
|
Project Partners: |
|
Department: |
Electrical Engineering |
Organisation: |
Victoria University of Manchester, The |
Scheme: |
Standard Research (Pre-FEC) |
Starts: |
01 November 1993 |
Ends: |
30 April 1996 |
Value (£): |
94,310
|
EPSRC Research Topic Classifications: |
Human Communication in ICT |
|
|
EPSRC Industrial Sector Classifications: |
|
Related Grants: |
|
Panel History: |
|
Summary on Grant Application Form |
To develop efficient mixed excitation source models for speech production. To employ such models and thus design high quality speech codecs operating in the 1.2 to 2.4 kbits/sec range. To examine the embedded variable bit rate characteristics of the codecs, within the 1.2 to 4 kbits/sec range. To investigate the codec performance characteristics under channel noise conditions.Progress:Work in this programme builds upon the relatively recently developed Manchester, low bit rate mixed excitation, speech production source-filter model (MSPM). The model exploits the long term signal correlation, present in voiced speech, using 'prototype' pitch related segments and achieves a highly compact representation of the excitation signal in a way which represents accurately the continuous evolution of the signal dynamics.In particular, work has been focused on the following issues: selection of domain employed to represent the pitch related prototype segments/excitation signals. efficient quantization of these prototype segments. development of improved mixed/multiband excitation models. advanced coding techniques of the filter parameters. development of improved, integrated voiced/unvoiced classification and accurate pitch estimation procedures.As a result new codecs of relatively low complexity have been designed which operate in the region of 1.2 to 2.8 kbits/sec. Recent informal subjective tests performed using a wide range of multispeaker/multi language input speech clearly indicated that the speech quality of the new Manchester system at 2.4 kbits/sec is equivalent to that obtained from conventional 4.8 kbits/sec CELP codecs.Work is now moving into exploiting the variable bit rate characteristics of the above scheme with respect to i) the statistics of the source and ii) the bit rate imposed by the traffic of an associated transmission network.
|
Key Findings |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
Potential use in non-academic contexts |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
Impacts |
Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
Summary |
|
Date Materialised |
|
|
Sectors submitted by the Researcher |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
Project URL: |
|
Further Information: |
|
Organisation Website: |
|