Adaptive speech rate conversion without extension of input data duration, using speech interval detection

• US 20010010037A1
• Filed: 02/12/2001
• Published: 07/26/2001
• Est. Priority Date: 04/30/1997
• Status: Active Grant

• Current Assignee: Nippon HOSA Kyoka
• Original Assignee: Nippon HOSA Kyoka
• Inventors: Imai, Atsushi, Takagi, Tohru, Seiyama, Nobumasa
• Granted Patent: US 6,374,213 B2
• Time in Patent Office: Days
• US Class Current: 704/210
• CPC Class Codes: G10L 2025/786 Adaptive threshold
• CPC Class Codes: G10L 25/78 Detection of presence or absence of voice signals switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10

Abstract

When a delivered speed of a listening speech (speech speed) is slowed down, a connection order generator (8) always monitors a data length of input speech, an output data length calculated previously by a conversion function concerning a preset scaling factor, and a data length of actual output speech in predetermined processing unit, then decides connection order not to cause inconsistency among them. The speech data and the connection data are connected without omission of speech information by controlling a speech data connector (9). When power of an input signal data is calculated to discriminate a speech interval and a non-speech interval, a threshold value for power is decided according to a maximum value of the power and difference between the maximum value and a minimum value.

Claims

1. A speech interval detecting method comprising the steps of:

calculating a frame power of an input signal data in unit of predetermined frame width at a predetermined time interval, and then holding a maximum value and a minimum value of the frame power within a past predetermined time period;

deciding a threshold value for power changed according to the maximum value being held and difference between the maximum value and the minimum value; and

comparing the threshold value with power of a current frame to decide whether or not the current frame belongs to a speech interval or a non-speech interval.

2. A speech interval detecting method set forth in

claim 1

, wherein, if the difference between the maximum value and the minimum value is less than a predetermined value, the threshold value is decided close to the maximum value rather than a case where the difference between the maximum value and the minimum value is more than the predetermined value.

3. A speech interval detecting device comprising:

a power calculator (32) for calculating a frame power of an input signal data in unit of predetermined frame width at a predetermined time interval;

an instantaneous power maximum value latch (33) for holding a maximum value of the frame power within a past predetermined time period;

an instantaneous power minimum value latch (34) for holding a minimum value of the frame power within the past predetermined time period;

a power threshold value decision portion (35) for deciding a threshold value for power changed according to the maximum value being held in the instantaneous power maximum value latch and difference between the maximum value and the minimum value being held in the instantaneous power minimum value latch; and

a discriminator (36) for comparing the threshold value obtained by the power threshold value decision portion with power of a current frame to decide whether or not the current frame belongs to a speech interval or a non-speech interval.

4. A speech interval detecting device set forth in

claim 3

, wherein, if the difference between the maximum value and the minimum value is less than a predetermined value, the power threshold value decision portion (35) decides the threshold value close to the maximum value rather than a case where the difference between the maximum value and the minimum value is more than the predetermined value.

5. A speech speed converting method comprising the steps of:

reducing an extension time of output data with respect to input data by any time period within the extension time when non-speech intervals appears in the output data obtained by extending/synthesizing the input data at any time-changing ratio and also a continued time of the non-speech intervals exceeds a predetermined threshold value.

6. A speech speed converting method set forth in

claim 5

, further comprising the steps of:

executing synthesizing processes while monitoring sequentially not to cause inconsistency an input data length, a target data length calculated by multiplying the input data length by any scaling factor, and an actual output data length when the input data are expanded/contracted and synthesized; and

holding precise time information of extension caused in the speech speed conversion against any time-changing extension/scaling factors not to cause omission of speech information in the speech interval.

7. A speech speed converting method set forth in

claim 5

, further comprising the steps of:

changing adaptively a remaining ratio of the non-speech interval according to a speech speed conversion factor, an amount of extension, and the like by reducing a part of the non-speech interval which exceeds a constant continued time when an amount of the extension from the input data length in the speech speed conversion is eliminated.

8. A speech speed converting method set forth in

claim 5

, further comprising the steps of:

measuring an amount of extension at a preset time interval while monitoring sequentially not to cause inconsistency among an input data length, a target data length calculated by multiplying the input data length by any scaling factor, and an actual output data length when the speech speed conversion is executed within the limited time range; and

changing the speech speed conversion factor adaptively according to the measuring result by increasing temporarily the speech speed conversion factor if an amount of the time difference is small but decreasing temporarily the speech speed conversion factor if an amount of the time difference is large.

9. A speech speed converting method set forth in

claim 5

, when a speech interval and a non-speech interval are discriminated, the method further comprising the steps of:

calculating a frame power of an input signal data in unit of predetermined frame width at a predetermined time interval, and then holding a maximum value and a minimum value of the frame power within a past predetermined time period;

deciding a threshold value for power changed according to the maximum value being held and difference between the maximum value and the minimum value; and

comparing the threshold value with power of a current frame to decide whether or not the current frame belongs to a speech interval or a non-speech interval.

10. A speech speed converting method set forth in

claim 9

, wherein, if the difference between the maximum value and the minimum value is less than a predetermined value, the threshold value is decided close to the maximum value rather than a case where the difference between the maximum value and the minimum value is more than the predetermined value.

11. A speech speed converting device comprising:

a split processing/connection data generating means for generating block data by splitting input data into block data, and then generating connection data based on respective block data; and

a connection processing means for deciding connection order of respective block data generated by the split processing/connection data generating means and connection data based on desired speech speed being input, and then connecting respective block data and the connection data to generate output data;

wherein the connection processing means reduces an extension time of output data with respect to input data by any time period within the extension time when non-speech intervals appears in the output data obtained by extending/synthesizing the input data at any time-changing ratio and also a continued time of the non-speech intervals exceeds a predetermined threshold value.

12. A speech speed converting device set forth in

claim 11

, wherein the connection processing means executes synthesizing processes while monitoring sequentially not to cause inconsistency an input data length, a target data length calculated by multiplying the input data length by any scaling factor, and an actual output data length when the input data are expanded/contracted and synthesized, and

holds precise time information of extension caused in the speech speed conversion against any time-changing extension/scaling factors not to cause omission of speech information in the speech interval.

13. A speech speed converting device set forth in

claim 11

, wherein the connection processing means changes adaptively a remaining ratio of the non-speech interval according to a speech speed conversion factor, an amount of extension, and the like by reducing a part of the non-speech interval which exceeds a constant continued time when an amount of the extension from the input data length in the speech speed conversion is eliminated.

14. A speech speed converting device set forth in

claim 11

, wherein the connection processing means measures an amount of extension at a preset time interval while monitoring sequentially not to cause inconsistency among an input data length, a target data length calculated by multiplying the input data length by any scaling factor, and an actual output data length when the speech speed conversion is executed within the limited time range, and changes the speech speed conversion factor adaptively according to the measuring result by increasing temporarily the speech speed conversion factor if an amount of the time difference is small but decreasing temporarily the speech speed conversion factor if an amount of the time difference is large.

15. A speech speed converting device set forth in

claim 11

, further comprising an analysis processing means for calculating a frame power of an input signal data in unit of predetermined frame width at a predetermined time interval, and then holding a maximum value and a minimum value of the frame power within a past predetermined time period,

deciding a threshold value for power changed according to the maximum value being held and difference between the maximum value and the minimum value, and

comparing the threshold value with power of a current frame to decide whether or not the current frame belongs to a speech interval or a non-speech interval.

16. A speech speed converting device set forth in

claim 15

, wherein, if the difference between the maximum value and the minimum value is less than a predetermined value, the analysis processing means decides the threshold value close to the maximum value rather than a case where the difference between the maximum value and the minimum value is more than the predetermined value.