EP0243009A1
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Speech recognition system using Markov models
Assignee: International Business Machines Corp
•
Filing: 1987-03-25
•
Granted: —
•
Expires: 2007-03-25
Abstract
Abstract The present invention relates to a speech recognition system of the type comprising a plurality of probabilistic finite state models each having a number of states and the ability of undergoing transitions from one state to another and producing a corresponding output representing a speech element, and means for defining for each model probability parameters each representing the probability that the model will undergo a transition from one predetermined state to another predetermined state and produce a corresponding output. Such a system can be used to recognise input speech data by initially dividing the input speech data into individual speech elements (4, 5, 6) and then applying the input speech elements to the models, and utilising the probability parameters of the models to recognise the input speech elements. According to the invention the speech recognition system is characterised in that it comprises training means (8) for supplying training speech data to the models in order to train the models and to define initial values for the probability parameters for each of the models, and adaptation means (9) for supplying adaptation speech data to the models in order to adapt the models and to define adapted values of the probability parameters for each of the models. The adapted values of the probability parameters are used to recognise the input speech elements (10).
Inventors
- 👤 Akihiro Kuroda
- 👤 Masafumi Nishimura
- 👤 Kazuhide Sugawara
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-10-27 04:20
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/06; G10L15/065; G10L15/07; G10L15/08; G10L15/14; G10L15/142; G10L15/144
Claims Analysis
Total Claims: 10
Claim 1
Independent Claim
194 words
A speech recognition system comprising a plurality of probabilistic finite state models each having a number of states and the ability of undergoing transitions from one state to another and producing a corresponding output representing a speech element, means for defining for each model probability parameters each representing the probability that said model will undergo a transition from one predetermined state to another-predetermined state and produce a corresponding output, means (4, 5, 6) for dividing input speech data into individual speech elements and for applying said input speech elements to said models, and means for utilising the probability parameters of said models to recognise said input speech elements, characterised in that said system comprises training means (8) for supplying training speech data to said models in order to train said models and to define initial values for said probability parameters for each of said models, adaptation means (9) for supplying adaptation speech data to said models in order to adapt said models and to define adapted values of said probability parameters for each of said models, and means (10) for utilising said adapted values of said probability parameters to recognise said input speech elements.
Claim 2
Dependent Claim
194 words
A speech recognition system comprising a plurality of probabilistic finite state models each having a number of states and the ability of undergoing transitions from one state to another and producing a corresponding output representing a speech element, means for defining for each model probability parameters each representing the probability that said model will undergo a transition from one predetermined state to another-predetermined state and produce a corresponding output, means (4, 5, 6) for dividing input speech data into individual speech elements and for applying said input speech elements to said models, and means for utilising the probability parameters of said models to recognise said input speech elements, characterised in that said system comprises training means (8) for supplying training speech data to said models in order to train said models and to define initial values for said probability parameters for each of said models, adaptation means (9) for supplying adaptation speech data to said models in order to adapt said models and to define adapted values of said probability parameters for each of said models, and means (10) for utilising said adapted values of said probability parameters to recognise said input speech elements.
Claim 3
Dependent Claim
36 words
A speech recognition system as claimed in claim 1 characterised in that said training speech data comprises a plurality of speech elements closely related to said input speech elements and each repeated a plurality of times.
Claim 4
Dependent Claim
37 words
A speech recognition system as claimed in either of the preceding claims characterised in that said adaptation speech data comprises a plurality of speech elements closely related to said input speech elements and each occurring only once.
Claim 5
Dependent Claim
36 words
A speech recognition system as claimed in any one of the preceding claims characterised in that said adaptation means comprises means for modifying the probability parameters of said models before supplying said adaptation speech data thereto.
Claim 6
Dependent Claim
172 words
A speech recognition method comprising defining, for a plurality of probabilistic finite state models each having a number of states and the ability of undergoing transitions from one state to another and producing a corresponding output representing a speech element, probability parameters each representing the probability that said model will undergo a transition from one predetermined state to another predetermined state and produce a corresponding output, dividing input speech data into individual speech elements and applying said input speech elements to said models, and utilising the probability parameters of said models to recognise said input speech elements, characterised in that said method comprises supplying training speech data to said models in order to train said models and to define initial values for said probability parameters for each of said models, supplying adaptation speech data to said models in order to adapt said models and to define adapted values of said probability parameters for each of said models, and utilising said adapted values of said probability parameters to recognise said input speech elements.
Claim 7
Dependent Claim
172 words
A speech recognition method comprising defining, for a plurality of probabilistic finite state models each having a number of states and the ability of undergoing transitions from one state to another and producing a corresponding output representing a speech element, probability parameters each representing the probability that said model will undergo a transition from one predetermined state to another predetermined state and produce a corresponding output, dividing input speech data into individual speech elements and applying said input speech elements to said models, and utilising the probability parameters of said models to recognise said input speech elements, characterised in that said method comprises supplying training speech data to said models in order to train said models and to define initial values for said probability parameters for each of said models, supplying adaptation speech data to said models in order to adapt said models and to define adapted values of said probability parameters for each of said models, and utilising said adapted values of said probability parameters to recognise said input speech elements.
Claim 8
Dependent Claim
40 words
A speech recognition method as claimed in claim 5 characterised in that said training speech data suuplied to said models comprises a plurality of speech elements closely related to said input speech elements and each repeated a plurality of times.
Claim 9
Dependent Claim
41 words
A speech recognition method as claimed in claims 5 or claim 6 characterised in that said adaptation speech data supplied to said models comprises a plurality of speech elements closely related to said input speech elements and each occurring only once.
Claim 10
Dependent Claim
34 words
A speech recognition method as claimed in any one of claims 5, 6 of 7 characterised in that it comprises modifying the probability parameters of said models before supplying said adaptation speech data thereto.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.