Information retrieving method and apparatus

• US 7,747,435 B2
• Filed: 03/15/2008
• Issued: 06/29/2010
• Est. Priority Date: 01/25/2002
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Toguri, Yasuhiro, Nishiguchi, Masayuki
• Primary Examiner(s): Vo; Huyen X.
• Application Number:
  US12/075,872
• Publication Number:
  US 20080177535A1
• Time in Patent Office: 836 Days
• Field of Search: 704/278, 704/222, 704/219, 704/220, 704/223, 704/230, 704/221, 704/229, 704/246, 704/245, 704/247, 704/249, 704/231, 704/237
• US Class Current: 704/246
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques

Abstract

A speaker of encoded speech data recorded in a semiconductor storage device in an IC recorder is to be retrieved easily. An information receiving unit 10 in a speaker retrieval apparatus 1 reads out the encoded speech data recorded in a semiconductor storage device 107 in an IC recorder 100. A speech decoding unit 12 decodes the encoded speech data. A speaker frequency detection unit 13 discriminates the speaker based on a feature of the speech waveform decoded to find the frequency of conversation (frequency of occurrence) of the speaker in a preset time interval. A speaker frequency graph displaying unit 14 displays the speaker frequency on a picture as a two-dimensional graph having time and the frequency as two axes. A speech reproducing unit 16 reads out the portion of the encoded speech data corresponding to a time position or a time range specified by a reproducing position input unit 15 based on this two-dimensional graph from the storage device 11 and decodes the read-out data to output the decoded data to a speech outputting unit 17.

Claims

1. An information retrieving apparatus for retrieving information on a speaker from encoded speech data, said apparatus comprising:

means for determining a speaker for each recognition interval based on a feature of said encoded speech data;

wherein the means for determining the speaker in each interval is comprised of means for extracting only LSP (Linear Spectrum Pair) information from the encoded speech data, decoding the extracted information, converting the decoded information into LPC (Linear Predictive Coding) cepstrum coefficients and evaluating vector quantization distortion; and

a speaker frequency graph outputting step of outputting and displaying the frequency of occurrence of each speaker detected for each of said intervals as two-dimensional graphical data having time and the frequency of occurrence as two axes on an output image, wherein the image provides a numeric percentage representative of the speaker frequency for the interval.

2. An information retrieving method for retrieving information on a speaker, from encoded speech data, said method comprising:

a speaker frequency detecting step of determining the speaker for each recognition interval based on a feature of said encoded speech data; and

wherein the speaker frequency detecting step for determining the speaker of the recording in each interval is comprised of extracting only LSP (Linear Spectrum Pair) information from the encoded speech data, decoding the extracted information, converting the decoded information into LPC (Linear Predictive Coding) cepstrum coefficients and evaluating vector quantization distortion;

a speaker frequency graph outputting step of outputting and displaying the frequency of occurrence of each speaker detected for each of said intervals as two-dimensional graphical data having time and the frequency of occurrence as two axes on an output image, wherein the image provides a numeric percentage representative of the speaker frequency for the interval.

3. The information retrieving method according to claim 2, wherein, in said speaker frequency graph outputting step, the frequency of occurrence of a plurality of speakers is output and displayed on said image.

4. The information retrieving method according to claim 2, wherein said encoded speech data has been recorded on a random accessible recording device or recording medium;

said method further comprising;

a position specifying step of specifying a time position on said two-dimensional graphical data displayed on said output image; and

a speech reproducing step of reading-in said time position specified and reproducing said encoded speech data from said time position.

5. The information retrieving method according to claim 4, wherein, in said position specifying step, a time range on said two-dimensional graphical data is specified;

and

wherein, in said speech reproducing step, said encoded speech data of the time range specified is reproduced.

6. The information retrieving method according to claim 4, further comprising:

a speaker specifying step of specifying a desired speaker, wherein, in said speech reproducing step, the portion of said encoded speech data in which the frequency of occurrence of said desired speaker specified in said speaker specifying step is not less than a preset threshold value is reproduced.

7. The information retrieving method according to claim 4, wherein said encoded speech data is recorded in a semiconductor storage device in an IC (Integrated Circuit) recorder.

8. An information retrieving apparatus for retrieving information on a speaker from encoded speech data, said apparatus comprising:

means for determining a speaker for each recognition interval based on a feature of said encoded speech data; and

wherein the means for determining the speaker in each interval is comprised of means for extracting only LSP (Linear Spectrum Pair) information from the encoded speech data) decoding the extracted information, converting the decoded information into LPC (Linear Predictive Coding) cepstrum coefficients and evaluating vector quantization distortion;

means for outputting step of outputting and displaying the frequency of occurrence of each speaker detected for each of said intervals as two-dimensional graphical data having time and the frequency of occurrence as two axes on an output image, wherein the image provides a numeric percentage representative of the speaker frequency for the interval.