JPH06289895A - Real-time speaking speed converting method - Google Patents

Abstract

PURPOSE:To convert a speech into an easy-to-hear speech which is slow on the whole without substantially extending a speaking time by properly reducing the speaking speed of a part which is considered as an important meaning part in a spoken speech and contrary increasing the speed of other parts through real-time processing. CONSTITUTION:This method is equipped with a speech input circuit 1, a CPU circuit 2, a PROM circuit 3, an input buffer circuit 4, a processing buffer circuit 5, a file circuit 6, a speech output circuit 7, and a bus 8. Then the circuit 1 inputs a speech (original speech) to be an object converted in speaking speed and detects variation in the intonation (pitch frequency) of the original speech through the real-time processing. Further, the speaking speed is varied according to a rule which increases the speaking speed for a high-intonation part and decreasing the speaking speed for a low part according to the indication, thereby converting the original speech into the excellent, easy-to-hear speech while maintaining the vocalization time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention captures original voice,
The present invention relates to a real-time speech speed conversion method for converting to slow-speed voice suitable for hearing of hearing impaired persons and elderly people.

SUMMARY OF THE INVENTION The present invention relates to a real-time speech speed conversion method for capturing original sound and converting it into slow-speed sound suitable for listening to the sound of hearing impaired persons, the elderly, and the like. Changes in voice intonation (pitch frequency) are detected by real-time processing, and based on this detection result, the speech speed is changed according to the rule that the speech speed is slowed in the high intonation portion and increased in the low intonation portion. By doing so, the original voice is converted into a good voice that is easy to hear while maintaining the utterance time of the original voice.

[0003]

2. Description of the Related Art Generally, when a listener deteriorates his / her listening ability such as a voice recognition critical speed (maximum speech speed at which voice can be accurately identified) due to aging or some kind of disorder, a normal voice or quick voice is used. The degree of distinction of the spoken voice is greatly reduced.

In the past, only a hearing aid was available as a hearing aid for people with such hearing impairment.

[0005]

The hearing aid described above is a device for compensating only the transfer characteristics of the outer ear and middle ear of the auditory system by simply improving the frequency characteristics and controlling the gain. There is a problem in that it is not possible to compensate for the deterioration in the ability of discriminating voice that is related to deterioration.

Therefore, as a method for solving such a problem, a technique for converting the speech speed while maintaining the quality of the original voice has been developed.

In this speech speed conversion technology, it is possible to make the speech much easier to hear, especially for the elderly and deaf persons, by uniformly slowing only the speech speed of the speech. The utterance time of the voice is inevitably extended by the operation. However, in broadcasting etc., the utterance time of the voice before decompression is uttered so as to be within the predetermined time. Therefore, if the voice is decompressed in this way, it may not be within the above time limit. Occurs. Further, in the case where audio and video are provided in synchronism with each other such as in a television, if only the audio is expanded, there is a time lag between the audio and the video, which may adversely affect listening. Occur.

Therefore, although a speech speed conversion technique considering such a temporal "deviation" is realized by off-line processing, a speech speed conversion technique considering a temporal "deviation" is developed. What can be realized by real-time processing has not been developed yet.

In view of the above-mentioned circumstances, the present invention solves the above-mentioned problems associated with the "deviation" in time. Therefore, in real-time processing, the speech speed of a portion considered to be significant in the uttered voice is considered. To provide a real-time speech speed conversion method capable of converting to a slow and easy-to-listen voice as a whole without substantially extending the utterance time by moderately slowing it and speeding up the other parts in reverse. It is an object.

[0010]

In order to achieve the above-mentioned object, the real-time speech speed conversion method according to the present invention, when slowing down the utterance speed (speaking speed) of a listening voice, divides sentences into sentences. The feature is that it is detected and judged as a pause section, and the speech speed is changed based on a predetermined rule divided into a portion within a predetermined time range from this pause section and a portion outside this range. I am trying.

[0011]

With the above structure, in the present invention, when the speed at which the received voice is uttered (speech speed) is slowed down, a pause between sentences is detected to determine a pause section, and a predetermined time is passed from this pause section. By changing the speech speed based on a predetermined rule that is divided into a portion within the range and a portion outside this range, while solving the problems associated with temporal "deviation", By slowing down the speech speed of the meaningful part of the uttered voice moderately and speeding up the other parts in the opposite direction, the utterance time is not extended substantially and the voice is slow and easy to hear as a whole in real time. Convert to.

[0012]

【Example】

<< Structure of Embodiment >> FIG. 1 is a block diagram showing an example of a real-time speech speed conversion apparatus to which an embodiment of the real-time speech speed conversion method according to the present invention is applied.

The real-time speech speed conversion apparatus shown in this figure comprises a voice input circuit 1, a CPU circuit 2, a PROM circuit 3, an input buffer circuit 4, a processing buffer circuit 5,
The file circuit 6, the voice output circuit 7, and the bus 8 are provided, and the voice input circuit 1 takes in the voice (original voice) that is the target of speech speed conversion, and the intonation (pitch frequency) of the original voice is processed in real time. Based on this detection result, the speech speed is changed in accordance with the rule that the speech speed is slowed down in the high intonation portion and increased in the low intonation portion, so that the utterance time of the original voice is maintained. The original voice is converted into a good voice that is easy to hear.

The voice input circuit 1 is a circuit having a general structure for inputting an original voice, for example, a microphone, a tone circuit, an analog-digital converter, a voice memory reproduction (recording).
A circuit, a voice storage medium (for example, an IC memory, a hard disk, a floppy disk or a VTR), an interface circuit, and the like are provided, and the voice that is the target of speech rate conversion is taken in and converted into a digital format voice signal. The data is supplied to the input buffer circuit 4 frame by frame based on an instruction from the CPU circuit 2.

The input buffer circuit 4 has a required capacity of RA.
This is a portion configured by M and the like and used as a work area of the CPU circuit 2. The audio signal output from the audio input circuit 1 is captured and stored, and the CPU
Based on the instruction from the circuit 2, the stored audio signal is transferred to the processing buffer circuit 5.

The processing buffer circuit 5 has a required capacity of RA.
This is a portion configured by M and the like and used as a work area of the CPU circuit 2. The audio signal output from the input buffer circuit 4 is captured and stored, and C
Based on an instruction from the PU circuit 2, the stored audio signal is transferred to the file circuit 6 or the like.

The file circuit 6 is composed of a voice storage medium such as an IC memory or a floppy disk in addition to the RAM, and is subjected to a voice signal expanded voiced section and a silent section shortening process according to the present invention. A signal for storing processed signals, and when the processed audio signal is output from the processing buffer circuit 5, the audio signal is captured and stored, and then stored based on an instruction from the CPU circuit 2. Is supplied to the audio output circuit 7.

The audio output circuit 7 is a circuit having a general structure for outputting the audio signal in the file circuit 6 to the outside, for example, an interface circuit, a digital-analog converter,
It is equipped with a speaker, a recording device (or broadcasting equipment), etc., and when an audio signal is output from the file circuit 6,
This is taken in and converted into voice and output to the outside.

Further, the CPU circuit 2 is a portion constituted by a one-chip microcomputer, etc.
Based on the program stored in the OM circuit 3, the control of the entire apparatus and various data processing are performed.

Further, the PROM circuit 3 is a portion used as a storage place for a program for defining the operation of the CPU circuit 2 and constant data used in various processes.
In response to a read command from the CPU circuit 2, the stored program or constant data is read and supplied to the CPU circuit 2.

<< Operation of the Embodiment >> Next, the operation of this embodiment will be described with reference to the block diagram of FIG. 1, the flowcharts of FIGS. 2 and 3, and the timing chart of FIG.

First, the CPU circuit 2 cuts out a voice signal which is first input to the voice input circuit 1 and processed, at a constant length called a frame, for example, every 3.3 ms, and transfers this to the input buffer circuit 4 for storage. Allow (Step ST
1).

Thereafter, the CPU circuit 2 processes the voice signal stored in the input buffer circuit 4 for each frame by a method such as an autocorrelation method or a zero-cross method, and outputs voiced, unvoiced, or silent voice for each frame. Make a decision. However, in this case, as a general rule, input sounds other than voiced and unvoiced sounds (for example, low-level noise and background sounds) produced by humans are processed as silence (step ST2).

Next, the CPU circuit 2 determines whether the voiced, unvoiced, and silent determination results for the current frame are the same as the voiced, unvoiced, and silent determination results for the previous frame (step ST3). ), If these are the same type, return to the above-described frame cutout process and repeat the same process, and if a different type, for example, the previous frame is a voiced section and the current frame is an unvoiced section, Audio signals determined to be the same type of section are transferred to the processing buffer circuit 5 and stored therein (step ST4).

As a result, as shown in FIG. 4, the voice input by the voice input circuit 1 is divided into the voiced section, the unvoiced section, and the silent section and stored in the processing buffer circuit 5.

After that, the CPU circuit 2 selects a pause section (i.e., a voice output section) from the voice signals stored in the processing buffer circuit 5 in a silence section having a section length of 250 ms or more among the sections determined to be the silence section. It is determined to be a breath sound breathing section), and a section between each pause section is determined to be a phrase section (section to be uttered in one breath) (step ST
5).

Then, the CPU circuit 2 performs the voiced section processing 10 shown in FIG. 3 for the section determined as the voiced section of each phrase section (step ST6).

In this voiced section processing, the CPU circuit 2 first determines whether or not the voiced section to be processed is the voiced section immediately after the pause section (step ST15). Three voiced sections (first voiced section, second voiced section, third voiced section) are extracted from the start point (Ph_st) of the phrase section, and among these pitch frequencies of the first voiced section to the third voiced section, The highest pitch frequency is set to the highest pitch frequency Pitch_max, and the expansion rate of the speech speed at the start point V_st of the first voiced section is set to "rs" (step ST1).
6).

Thereafter, the CPU circuit 2 causes the voice signal to be processed to have a preset time T from the start point V_st of the first voiced section (2000 m in this embodiment).
s) has elapsed (step ST1
7) If the time T has not elapsed, the expansion rate of the speech speed is changed from "rs" to "re" using an appropriate decreasing function set in advance, for example, the cosine function f (t) shown in the following equation. It is changed (step ST18).

F (t) = re + (1/2) · (rs−re) · {cosπ · (t−V_st) /T+1.0} (1) However, t: t = V_st to V_st + T At this time, no processing is performed on the silent section and the unvoiced section in this range.

If the voice signal to be processed has passed the preset time T from the start point V_st of the first voiced section (step ST17), the CPU circuit 2 becomes the process target. Section containing the audio signal
It is determined whether the average pitch frequency Pitch (n) in the voice section) (where n ≧ k) satisfies the following equation (step ST19).

Pitch (n)> Pitch_max × Th2 (2) where Th2: threshold value, and in this embodiment, Th
2 = 0.7.

If the nth voice section satisfies the above expression (2), the CPU circuit 2 sets the start point of this nth voice section to "V2_st" (step ST20) and determines whether it is within the above-mentioned period. Determination process and decreasing function f
(T) is used to expand the voiced section, and the expansion rate of the speech speed is changed from "rs-Th3" to "re" within the range from the start point V2_st to the period T (step ST18).

In this case, in this embodiment, the threshold value Th
The value of 3 is set to "0.1".

If the equation (2) is not satisfied (step ST19), the CPU circuit 2 keeps the expansion rate re of the voiced section, that is, the fastest speech speed (step ST21).

Thereafter, the CPU circuit 2 repeats the above-described processing for the voice signal in the voiced section every time the voiced section is detected until the next pause section.

After this processing is completed, the CPU circuit 2 transfers the speech-rate-converted voice signal in the processing buffer to the file circuit 6 for storage and also clears the processing buffer circuit 5 (step ST7). ).

Further, in the above-mentioned identification section processing (step ST5), if the section to be processed is determined to be the unvoiced section, the CPU circuit 2 transfers the voice signal of this section from the processing buffer circuit 5 to the file circuit 6. Then, the processing buffer circuit 5 is cleared (step ST
8).

Further, in the above-described identification section processing (step ST5), if the section to be processed is determined to be a silent section, the CPU circuit 2 determines whether this section is a pause section (step ST9), and the pause section. When it is, it is determined that the sentence is a break (a phrase) between sentences and the sentence is shortened to the shortest without any discomfort in hearing. Therefore, a preset algorithm is shortened to shorten the silent section ( Step ST10).

Further, in the above-described identification section processing (step ST5), even if it is determined that there is no sound section, if it is not a pause section (step ST9), the CPU circuit 2 skips this shortening processing.

After that, the CPU circuit 2 outputs the signal of the processed silent section in the processing buffer circuit 5 to the file circuit 6.
Then, the processing buffer circuit 5 is cleared (step ST11).

Thereafter, the CPU circuit 2 repeats the above-described processing until there are no audio signals to be processed (step ST12).

In parallel with the above processing, the CPU circuit 2 transfers the processed audio signal stored in the file circuit 6 to the audio output circuit 7 and outputs it as audio.

<< Experimental example >> Then, when the voice including the actual news voice shown in Table 1 is processed by the above-mentioned real-time speech speed conversion device, for (1) -1 to (1) -6 in the sentence, Recognizing each break as a pause and slowing down the speech speed at the start point of each phrase, (1) -1, (1)
-4 was able to slow down the speech speed in the latter half (underlined part).

[0045]

[Table 1] Then, in this experiment, the extension rate of the speech speed is set to "rs =
When 1.25 ”and“ re = 0.9 ”, the original voice having a duration of 136 seconds can be changed to a voice having a duration of 136 seconds, and the speech speed is“ rs = re = 1.3 ”. The absorption rate α is 100% (unabsorbed time 0.0
Seconds).

In this case, the absorption rate α is a value represented by the following equation.

Α = {(T1−T2) / T1} · 100 (3) However, T1: extension time when the voice speed is uniform T2: extension time when the voice speed is changed Therefore, this embodiment By using, it is possible to effectively reduce the silent intervals between sentences, thereby converting the original voice into a slow voice in real time without extending the overall time length.

As described above, in this embodiment, the voice input circuit 1 takes in the voice (original voice) to be converted into the speech speed and detects the change in the intonation (pitch frequency) of the original voice in real time processing. Based on this detection result, in parts with high intonation,
By changing the speaking speed according to the rule of slowing the speaking speed and increasing the speaking speed in the low part, it is possible to convert the original speech into a good voice that is easy to hear while maintaining the utterance time of the original speech. Speaking speed can be moderately slowed down in the important part of the uttered voice, and speeded up in the other parts, which is slower and easier to hear as a whole without substantially extending the speaking time. Can be converted to voice.

[0049]

As described above, according to the present invention, in order to solve the problem caused by the "deviation" in time, in real-time processing, the part of the speech that is considered to be the important part in the uttered speech is talked about. By slowing the speed moderately and accelerating the other parts in the opposite direction, it is possible to convert the speech into a slow and easy-to-listen voice as a whole without substantially extending the utterance time.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a real-time voice speed conversion apparatus to which an embodiment of a real-time voice speed conversion method according to the present invention is applied.

FIG. 2 is a main flowchart showing an operation example of the real-time speech speed conversion device shown in FIG.

3 is a flowchart showing an example of a voiced segment processing routine showing an operation example of the real-time speech speed conversion apparatus shown in FIG.

FIG. 4 is a timing diagram showing an operation example of the real-time speech speed conversion device shown in FIG.

[Explanation of symbols]

 1 audio input circuit 2 CPU circuit 3 PROM circuit 4 input buffer circuit 5 processing buffer circuit 6 file circuit 7 audio output circuit 8 bus

Claims (1)

[Claims] 1. When slowing down the speed at which the listening voice is uttered (speech speed), a boundary between sentences is detected to determine a pause section, and a portion within a predetermined time range from this pause section, A real-time voice speed conversion method, characterized in that the voice speed is changed based on a predetermined rule divided into parts outside this range.