Voice quality change portion locating apparatus

• US 7,809,572 B2
• Filed: 06/05/2006
• Issued: 10/05/2010
• Est. Priority Date: 07/20/2005
• Status: Expired due to Fees

• Current Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Inventors: Yamagami, Katsuyoshi, Adachi, Shinobu, Kato, Yumiko
• Primary Examiner(s): Hudspeth; David R
• Assistant Examiner(s): Guerra-Erazo; Edgar
• Application Number:
  US11/996,234
• Publication Number:
  US 20090259475A1
• Time in Patent Office: 1,583 Days
• Field of Search: 704/260, 704/267, 704/258, 704/264, 704/220, 704/200
• US Class Current: 704/260
• CPC Class Codes: G10L 13/10 Prosody rules derived from text; Stress or intonation

Abstract

A text edit apparatus which presents, based on language analysis information regarding a text, a portion of the text where voice quality may change when the text is read aloud has advantages of predicting likelihood of the voice quality change and judging whether or not the voice quality change will occur. The apparatus includes: a voice quality change estimation unit (103) which estimates the likelihood of the voice quality change which occurs when the text is read aloud, for each predetermined unit which is an input symbol sequence of the text including at least one phonologic sequence, based on language analysis information which is a symbol sequence of a result of language analysis including a phonologic sequence corresponding to the text; a voice quality change portion judgment unit (105) which locates a portion of the text where the voice quality change is likely to occur, based on the language analysis information and a result of the estimation performed by the voice quality change estimation unit (103); and a display unit (108) which presents the user the portion which is located by the voice quality change portion judgment unit (105) as where the voice quality change is likely to occur.

Claims

1. A voice quality change portion locating apparatus which locates, based on language analysis information regarding a text, a portion of the text where voice quality may change when the text is read aloud, said apparatus comprising:

a storage unit in which a rule is stored, the rule being used for judging likelihood of the voice quality change based on phoneme information and prosody information;

a voice quality change estimation unit operable to estimate the likelihood of the voice quality change which occurs when the text is read aloud, for each predetermined unit of an input symbol sequence including at least one phonologic sequence, based on (i-1) phoneme information and (i-2) prosody information which are included in the language analysis information that is a symbol sequence of a result of language analysis including a phonologic sequence corresponding to the text, and (ii) the rule; and

a voice quality change portion locating unit operable to locate a portion of the text where the voice quality change is likely to occur, based on the language analysis information and a result of the estimation performed by said voice quality change estimation unit.

2. The voice quality change portion locating apparatus according to claim 1,

wherein the rule is an estimation model of the voice quality change, the estimation model being generated by performing analysis and statistical learning on voice of a user.

3. The voice quality change portion locating apparatus according to claim 1,

wherein said voice quality change estimation unit is operable to estimate the likelihood of the voice quality change for the each predetermined unit of the language analysis information, based on each of a plurality of utterance modes of a user, using a plurality of estimation models which are set for respective kinds of voice quality changes and generated by performing analysis and statistical learning on respective voices of the plurality of utterance modes.

4. The voice quality change portion locating apparatus according to claim 1,

wherein said voice quality change estimation unit is operable to (i) select an estimation model corresponding to each of a plurality of users, from among a plurality of estimation models for the voice quality change which are generated by performing analysis and statistical learning on respective voices of the plurality of users, and (ii) estimate the likelihood of the voice quality change for the each predetermined unit of the language analysis information, using the selected estimation model.

5. The voice quality change portion locating apparatus according to claim 1, further comprising:

an alternative expression storage unit in which an alternative expression for a language expression is stored; and

an alternative expression presentation unit operable to (i) search said alternative expression storage unit for an alternative expression for the portion of the text where the voice quality change is likely to occur, and (ii) present the alternative expression.

6. The voice quality change portion locating apparatus according to claim 1, further comprising:

an alternative expression storage unit in which an alternative expression for a language expression is stored; and

a voice quality change portion replacement unit operable to (i) search said alternative expression storage unit for an alternative expression for the portion of the text which is located by said voice quality change locating unit as where the voice quality change is likely to occur, and (ii) replace the portion by the alternative expression.

7. The voice quality change portion locating apparatus according to claim 6, further comprising

a voice synthesis unit operable to generate voice by which the text in which the portion is replaced by the alternative expression by said voice quality change portion replacement unit is read aloud.

8. The voice quality change portion locating apparatus according to claim 1, further comprising

a voice quality change portion presentation unit operable to present a user the portion of the text which is located by said voice quality change locating unit as where the voice quality change is likely to occur.

9. The voice quality change portion locating apparatus according to claim 1, further comprising

a language analysis unit operable to (i) perform the language analysis on the text, and (ii) output the language analysis information which is the symbol sequence of the result of the language analysis including the phonologic sequence.

10. The voice quality change portion locating apparatus according to claim 1,

wherein said voice quality change estimation unit is operable to estimate the likelihood of the voice quality change for the each predetermined unit, using, as an input, at least a kind of a phoneme, the number of moras in an accent phrase, and an accent position among the language analysis information.

11. The voice quality change portion locating apparatus according to claim 1, further comprising

an elapsed-time calculation unit operable to calculate an elapsed time which is a time period of reading from a beginning of the text to a predetermined position of the text, based on speech rate information indicating a speed at which a user reads the text aloud,

wherein said voice quality change estimation unit is further operable to estimate the likelihood of the voice quality change for the each predetermined unit, by taking the elapsed time into account.

12. The voice quality change portion locating apparatus according to claim 1, further comprising

a voice quality change ratio judgment unit operable to judge a ratio of (i) the portion which is located by said voice quality change locating unit as where the voice quality change is likely to occur, to (ii) all or a part of the text.

13. The voice quality change portion locating apparatus according to claim 1, further comprising:

a voice recognition unit operable to recognize voice by which a user reads the text aloud;

a voice analysis unit operable to analyze an occurrence degree of the voice quality change, for each predetermined unit which includes each phoneme unit of the voice of the user, based on a result of the recognition performed by said voice recognition unit; and

a text evaluation unit operable to compare (i) the portion of the text which is located by said voice quality change locating unit as where the voice quality change is likely to occur to (ii) a portion where the voice quality change has actually occurred in the voice of the user, based on (a) the portion of the text where the voice quality change is likely to occur and (b) a result of the analysis performed by said voice analysis unit.

14. The voice quality change portion locating apparatus according to claim 1,

wherein the rule is a phoneme-based voice quality change table in which a level of the likelihood of the voice quality change is represented for the each phoneme by the numeric value, and

said voice quality change estimation unit is operable to estimate the likelihood of the voice quality change for the each predetermined unit of the language analysis information, based on the numeric value which is allocated to each phoneme included in the predetermined unit, with reference to the phoneme-based voice quality change table.

15. A voice quality change portion locating apparatus which locates, based on language analysis information regarding a text, a portion of the text where voice quality may change when the text is read aloud, said apparatus comprising

a voice quality change portion locating unit operable to (i) locate a mora in the text as a portion where the voice quality change is likely to occur, the mora being one of (1) a mora, whose consonant is “

b”

that is a bilabial and plosive sound, and which is a third mora in an accent phrase, (2) a mora, whose consonant is “

m”

that is a bilabial and nasalized sound, and which is the third mora in the accent phrase, (3) a mora, whose consonant is “

n”

that is an alveolar and nasalized sound, and which is a first mora in the accent phrase, and (4) a mora, whose consonant is “

d”

that is an alveolar and plosive sound, and which is the first mora in the accent phrase, and also (ii) locate a mora in the text as a portion where the voice quality change is likely to occur, the mora being one of (5) a mora, whose consonant is “

h”

that is a guttural and unvoiced fricative, and which is one of the first mora and the third mora in the accent phrase, (6) a mora, whose consonant is “

t”

that is an alveolar and unvoiced plosive sound, and which is a fourth mora in the accent phrase, (7) a mora, whose consonant is “

k”

that is a velar and unvoiced plosive sound, and which is a fifth mora in the accent phrase, and (8) a mora, whose consonant is “

s”

that is a dental and unvoiced fricative, and which is a sixth mora in the accent phrase.

16. A voice quality change portion locating method of locating, based on language analysis information regarding a text, a portion of the text where voice quality may change when the text is read aloud, said method comprising steps of:

estimating likelihood of the voice quality change which occurs when the text is read aloud, for each predetermined unit of an input symbol sequence including at least one phonologic sequence, based on (i) a rule which is used for judging likelihood of the voice quality change according to phoneme information and prosody information, the phoneme information and prosody information being included in the language analysis information that is a symbol sequence of a result of language analysis including a phonologic sequence corresponding to the text, and (ii-1) the phoneme information and (ii-2) the prosody information; and

locating a portion of the text where the voice quality change is likely to occur, based on the language analysis information and a result of said estimating.

17. A non-transitory computer-readable medium encoded with computer executable instructions for locating, based on language analysis information regarding a text, a portion of the text where voice quality may change when the text is read aloud, said computer executable instructions causing a computer to execute steps of:

estimating likelihood of the voice quality change which occurs when the text is read aloud, for each predetermined unit of an input symbol sequence including at least one phonologic sequence, based on (i) a rule which is used for judging likelihood of the voice quality change according to phoneme information and prosody information, the phoneme information and prosody information being included in the language analysis information that is a symbol sequence of a result of language analysis including a phonologic sequence corresponding to the text, and (ii-1) the phoneme information and (ii-2) the prosody information; and

locating a portion of the text where the voice quality change is likely to occur, based on the language analysis information and a result of said estimating.