Process for evaluating speech quality in speech synthesis

• US 5,664,050 A
• Filed: 03/21/1996
• Issued: 09/02/1997
• Est. Priority Date: 06/02/1993
• Status: Expired due to Term

• Current Assignee: Intellectual Ventures I LLC (Intellectual Ventures LLC)
• Original Assignee: Telia AB (Government of Norway)
• Inventors: Lyberg, Bertil
• Primary Examiner(s): Tung, Kee M.
• Application Number:
  US08/622,039
• Time in Patent Office: 530 Days
• Field of Search: 395/2.4, 395/2.45, 395/2.53, 395/2.55, 395/2.6, 395/2.67
• US Class Current: 704/251
• CPC Class Codes: G10L 13/04 Details of speech synthesis systems, e.g. synthesiser structure or memory management
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

The present invention relates to a process for evaluating speech quality in speech synthesizers with the aid of a speech recognition system. The recognition system is programmed using a number of persons. It then receives synthetic or natural speech from speech synthesizers or persons respectively, displaying differing speech quality. The speech recognition system determines a level of recognition for each received speech quality. In order to evaluate the speech quality in a speech synthesizer, speech is received from the synthesizer by the speech recognition system, is allocated a level of recognition and is ranked in comparison to the levels of recognition for previously received speech.

Claims

1. A process for evaluating speech quality in synthetic speech generated from a speech synthesizer in a speech recognition system programmed using speech from a training group, comprising the steps of:

receiving synthetic speech from at least one speech synthesizer;

receiving natural speech from a plurality of human sources;

determining a level of recognition for each received speech from the at least one speech synthesizer and plurality of human sources; and

evaluating the synthetic speech from the at least one speech synthesizer by allocating a level of recognition and a rank of the synthetic speech from the at least one speech synthesizer in comparison to the natural speech from the plurality of human sources.

2. A system for evaluating speech quality in synthetic speech generated from a speech synthesizer comprising:

at least one speech synthesizer for generating synthetic speech;

a speech recognition system programmed using speech from a training group and for receiving the synthetic speech from the at least one speech synthesizer and natural speech from a plurality of human sources;

wherein a level of recognition for each received speech from the at least one speech synthesizer and the plurality of human sources is determined; and

wherein the synthetic speech from the at least one speech synthesizer is evaluated by allocating a level of recognition and a rank of the synthetic speech from the at least speech synthesizer in comparison to the natural speech from the plurality of human sources.

3. A system for evaluating speech quality in synthetic speech generated from a speech synthesizer comprising:

at least one speech synthesizer for generating synthetic speech;

a speech recognition system programmed using speech from a training group of a first plurality of human sources and for receiving the synthetic speech from the at least one speech synthesizer and natural speech from a second plurality of human sources;

wherein a level of recognition for each received speech from the at least one speech synthesizer and the second plurality of human sources is determined; and

wherein the synthetic speech from the at least one speech synthesizer is evaluated by allocating a level of recognition and a rank of the synthetic speech from the at least speech synthesizer in comparison to the natural speech from the second plurality of human sources.

4. The system according to claim 3, wherein the speech recognition system establishes a level of recognition of the training group of the first plurality of human sources at 100%.