Apparatus and method for normalizing an input speech signal

• US 6,275,795 B1
• Filed: 01/08/1999
• Issued: 08/14/2001
• Est. Priority Date: 09/26/1994
• Status: Expired due to Term

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Tzirkel-Hancock, Eli
• Primary Examiner(s): Tsang, Fan
• Assistant Examiner(s): Opsasnick, Michael N.
• Application Number:
  US09/227,711
• Time in Patent Office: 949 Days
• Field of Search: 704/201, 704/211, 704/214
• US Class Current: 704/214
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 25/09 the extracted parameters being zero crossing rates
• CPC Class Codes: G10L 25/15 the extracted parameters being formant information
• CPC Class Codes: G10L 25/93 Discriminating between voiced and unvoiced parts of speech signals G10L25/90 takes precedence

Abstract

In an apparatus for extracting information from an input speech signal, a preprocessor, a buffer, a segmenter, an acoustic classifier and a feature extractor are provided. The preprocessor generates formant related information for consecutive time frames of the input speech signal. This formant related information is fed into the buffer, which can store signals representative of a plurality of frames. The segmenter monitors the signals representative of the incoming frames and identifies segments in the input speech signal during which variations in the formant related information remain within prespecified limits. The acoustic classifier then determines classification information for each segment identified by the segmenter, based on acoustic classes found in training data. The feature estimator then determines, for each segment, the information required, based on the input speech signal during that segment, training data and the classification information determined by the acoustic classifier.

Claims

1. An apparatus for normalising an input speech signal, the apparatus comprising:

dividing means arranged to divide the input speech signal into a succession of frames;

level determining means arranged to determine a level of the input speech signal during each said frame;

normalising means arranged to normalise the level of the speech signal in each frame;

detection means arranged to detect a voiced signal in the input speech signal; and

control means arranged to adjust said normalising means when said detection means detects a voiced signal.

2. An apparatus according to claim 1, wherein said level is a power level of the input speech signal.

3. An apparatus according to claim 1, wherein said detection means comprises zero crossing detection means arranged to detect when the input speech signal crosses a zero level, and wherein said control means is arranged to adjust said normalising means when said zero crossing detector indicates that the frequency at which the input speech signal crosses the zero level falls below a threshold.

4. An apparatus according to claim 1, wherein said control means is arranged to adjust said normalising means according to the following equation:

5. An apparatus according to claim 4, wherein the values gⁱ, gⁱ⁻

1 and M in the equation are logarithmically scaled values.

6. A method for normalising an input speech signal, the method comprising:

the step of dividing the input speech signal into a succession of frames;

the step of determining a level of the input speech signal during each said frame;

the step of normalising the level of the speech signal in each frame;

the step of detecting a voiced signal in the input speech signal;

the step of adjusting said step of normalising when said detecting step detects a voiced signal in the input speech signal.

7. A method according to claim 6, wherein said level is a power level of the input speech signal.

8. A method according to claim 6, wherein said step of detecting comprises the step of detecting when the input speech signal crosses a zero level and wherein said adjustment is performed when said detecting step indicates that the frequency at which the input speech signal crosses the zero level falls below a threshold.

9. A method according to claim 6, wherein said normalising step is adjusted according to the following equation:

10. A method according to claim 9, wherein the values gⁱ, gⁱ⁻

1 and M in the equation are logarithmically scaled values.

11. An apparatus for normalising an input speech signal, the apparatus comprising:

dividing means for dividing the input speech signal into a succession of frames;

level determining means for determining a level of the input speech signal during each of said frames;

normalising means for normalising the level of the speech signal in each frame using a normalisation factor;

detection means for detecting a portion of the input speech signal having periodic characteristics indicative of a vocalized sound; and

control means for adjusting said normalisation factor using the speech signal within the detected portion.