EP 1649450 A1 20060426 - SPEECH RECOGNITION METHOD, AND COMMUNICATION DEVICE
Title (en)
SPEECH RECOGNITION METHOD, AND COMMUNICATION DEVICE
Title (de)
VERFAHREN ZUR SPRACHERKENNUNG UND KOMMUNIKATIONSGER T
Title (fr)
PROCEDE DE RECONNAISSANCE VOCALE ET APPAREIL DE COMMUNICATION
Publication
Application
Priority
- EP 2004050687 W 20040504
- DE 10334400 A 20030728
Abstract (en)
[origin: WO2005013261A1] Disclosed is a method for speaker-dependent speech recognition, comprising the following steps: a speech signal (SS) encompassing a voice command is detected; the speech signal (SS) is divided (F) into time frames; the speech signal is characterized in each detected time frame by forming a corresponding property vector (F_IS); a property vector sequence is formed from one or several property vectors (F_IS); the property vector sequence is stored; the property vector sequence is allocated to a sequence of model vectors in a speech resource (HMM-L) comprising a plurality of model vectors; a piece of allocation information indicating allocation of the property vector sequence to a sequence of model vectors is stored; and the voice command allocated to the model vector sequence is identified.
IPC 1-7
IPC 8 full level
G10L 15/06 (2013.01); G10L 15/14 (2006.01); G10L 15/22 (2006.01)
CPC (source: EP US)
G10L 15/063 (2013.01 - EP US); G10L 15/142 (2013.01 - EP US); G10L 2015/223 (2013.01 - EP)
Citation (search report)
See references of WO 2005013261A1
Designated contracting state (EPC)
DE ES FR GB IT
DOCDB simple family (publication)
WO 2005013261 A1 20050210; CN 1856820 A 20061101; DE 10334400 A1 20050224; EP 1649450 A1 20060426; JP 2007500367 A 20070111; US 2007112568 A1 20070517; US 7630878 B2 20091208
DOCDB simple family (application)
EP 2004050687 W 20040504; CN 200480027941 A 20040504; DE 10334400 A 20030728; EP 04741506 A 20040504; JP 2006521560 A 20040504; US 56629304 A 20040504