Loquendo Releases New Acoustic Models for Loquendo ASR

Loquendo released Loquendo ASR with two developments: acoustic models for the Aurora Distributed Speech Recognition (DSR) now in U.K. English and German, and acoustic models for the French language.

Loquendo developed acoustic models in U.K. English and German for the ETSI Aurora Distributed Speech Recognition that are compliant with the ETSI ES 202 050 and ES 202 212 specifications. Loquendo ASR is able to process front-end signal feature streams for DSR/Aurora systems.

Developments have also been made for the French language acoustic models: these acoustic-phonetic models are based on an optimized version of the set of phonetic units. The speech corpora used for the training of the acoustic models were processed by a tool able to discard recorded material displaying defects (e.g. poor transcription, distortion, background noise).