JPH0481198B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of the Invention] The present invention relates to a storage device for digital voice signals of a voice response device used in an information processing device. In particular, the present invention relates to an audio signal storage device that breaks down a sentence into words and phrases and writes each phrase individually into a storage device.

[Description of prior art]

Some voice response devices combine pre-registered words and phrases and reproduce them as a series of sentences. An appropriate amount of unvoiced parts are provided at the leading and trailing edges of these words and phrases, so that the sounds will be natural when combined and reproduced as a series of sentences.

Conventionally, in devices for writing such words and phrases, the words and phrases are written to the storage device in a form with sufficient silent parts added to the leading and trailing edges, and while being monitored, the silent parts at the leading and trailing edges are adjusted as appropriate. The cutting process was repeated until the length was reached. However, with this conventional method, the silent parts are cut out manually and rely on auditory sense, so it takes skill and a lot of time to cut out the silent parts appropriately so that the sound sounds natural when combined and played back. There was a drawback that it required In addition, since it was done by hand, the length of the silent portions left on the leading and trailing edges varied, making it difficult to obtain products of consistent quality.

[Purpose of the invention]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a digital audio signal storage device that eliminates the above-mentioned drawbacks and can reliably ensure a specified amount of silence at the leading and trailing edges of the audio signal.

[Features of the invention]

The present invention first determines the beginning and end positions of the voiced parts of a word or phrase when cutting out unnecessary silent parts added to the leading and trailing edges of the word or phrase,
Next, the silent part is cut out so that a specified amount of silent part is left before and after the sound part.

That is, the present invention includes means for converting input analog audio into digital audio signals and writing them into the storage device in frame units, and controlling writing to the storage device by external writing start and stop instructions. means for storing a write start address and stop address of the digital audio signal; and means for determining a frame as a silent frame when a predetermined number of frames of the written digital audio signal include a predetermined number or more of patterns indicating silence. and an input means for instructing to leave N frames at the leading edge of the digital audio signal and M frames at the trailing edge of the digital audio signal, and determines the silent frame frame by frame sequentially from the start address of the digital audio signal, VS the first frame position that is not determined to be a frame, and 1 from the stop address
Determine silent frames frame by frame, and VE the first frame position that is not determined to be a silent frame.
The system is characterized in that the VS-N frame to the VE+M frame are valid voice phrases.

Note that a frame here refers to a unit of an audio signal when a digital audio signal is divided into fixed byte units.

[Explanation based on examples]

Embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a digital audio signal storage device according to an embodiment of the present invention. In FIG. 1, an analog audio signal _S1, which is a word or phrase inputted from the outside, is sent to an analog-to-digital converter 1.
connected to. The digital audio signal converted from the analog-to-digital converter 1 is provided to the audio memory unit 2 in units of frames.

Here, the feature of the present invention is the storage portion of the digital audio signal, which is surrounded by a dashed line. That is, a write start signal _S2 is input from the outside to the control unit 3, and control signals such as a write start address and an increment instruction are connected from the control unit 3 to the counter ₄ via the signal line l1.
A write stop signal _S3 is input to the control section 3, and a control signal instructing to stop incrementing is connected from the control section 3 to the counter ₄ via the signal line l1. The output signal from counter 4 is connected to signal line l ₂
It is connected to the audio memory unit 2 via the audio memory section 2, and digital audio signals are stored therein. Further, the output signal from the counter 4 is branched and connected to the control section 3. The digital audio signal stored from the audio memory section 2 is connected to the silence pattern detection section 5, and the silence pattern detection signal is connected to the control section 3. The control unit 3 counts the silent pattern detection signal, determines a silent frame, and determines the first sound frame and the last sound frame. External silent frame designation signal S ₄
is input to the control unit 3, and a control signal is sent from the control unit 3 to the counter ₄ via the signal line l1 to leave a designated number of silent frames at the leading edge of the first sound frame and the trailing edge of the last sound frame. Connected. An output signal from the counter 4 is connected to the audio memory section 2 and the control section 3 via a signal line _l2 . A control signal is connected from the control unit 3 to an external storage device, and an audio phrase in which a specified number of silent frames are added to the leading and trailing edges of a sound frame is connected from the audio memory unit 2 to the external storage device. .

The operation of the digital signal storage device having such a configuration will be explained. Figure 2 shows the voice memory section 2.
FIG. 3 is a memory map diagram of digital audio signals stored in the . In FIG. 1, an analog audio signal _S1 , which is a word or phrase inputted from the outside, is converted into a digital audio signal by an analog-to-digital converter 1, and is stored in an audio memory 2 frame by frame. At this time, the external write start signal _S2
The control unit 3 determines the start address of the audio memory unit 2, sets the write start address in the counter 4, and thereafter sends an increment instruction to the counter 4 at a constant cycle until a write stop signal _S3 is received from the outside. . If there is a write stop signal _S3 from outside, the control section 3 stops sending out the above-mentioned increment instruction and ends the write operation to the audio memory section 2.

The storage state of the audio memory section 2 at this time will be explained with reference to FIG. START shown in FIG. 2a is a write start address, and STOP is a write stop address. Further, the shaded area indicates a digital audio signal with sound, and the blank area indicates a digital audio signal with no sound.

To explain how to determine silent frames,
A predetermined number of frames are read out from the write start address START of the digital audio signal stored in the audio memory section 2 shown in FIG. 1 and supplied to the silent pattern detection section 5. The control unit 3 counts the silence pattern detection signal output from the silence pattern detection unit 5 and compares it with a predetermined set number. If the count result is large, this predetermined number of frames are determined to be silent frames. Similarly, next time, from the write stop address STOP, the number of silent patterns in a predetermined number of bytes is determined, and a silent frame is determined. By sequentially performing the above operations and finding frames that are determined to be non-silent frames, the first and last frame positions of the sound portion are determined.

In the thus determined sound signal section, the first sound frame position is moved to the leading edge frame position, and the last frame position is moved to the trailing edge frame position by the number of frames specified by the external silent frame designation signal _S4 . By moving to the frame position, silent frames are added to the leading edge and the trailing edge by the number of externally specified frames to generate the target audio phrase.

This will be explained in more detail using FIG. 2a. First, in this example, a voice phrase is created in which a silent frame is created when the silent pattern is 80% or more, and one silent frame is left at each of the leading and trailing edges of the sound part. A silent frame is determined from frame A, which is the writing start point. Frame A and frame B are completely silent and have a silent pattern of 100%, so they are considered silent frames.
In the following frame C, 1/3 is a silent pattern and 80%
Since the frame is full, it is determined that it is not a silent frame. Similarly, if it is determined that the write stop point frame z goes up from frame Y to frame X, frame X
Since the silence pattern of is less than 80%, it is determined that the frame is not a silent frame. From this, it can be seen that frame C and frame X are the sound frames at both ends. Next, since it is specified from the outside to leave a silent frame one frame at a time, frame B, which is one frame before frame C, and frame X
Figure 2b, which includes frame Y one frame later than
As shown in the figure, frames B to Y are determined to be valid voice phrases.

The audio phrase created as above is
The control unit 3 shown in FIG. 1 reads out the audio data from the audio memory unit 2 and sends it to an external storage device.

〔Effect of the invention〕

As explained above, by providing a control section, a counter section, and a silence pattern detection section, the present invention automatically enables effective use when writing voice phrases such as words and phrases used in a voice response device into a storage device. It is advantageous to determine the leading edge and trailing edge of the sound signal, and to add a specified amount of silence signal to the leading edge and trailing edge of the sound signal. Therefore, since the lengths of the silent parts at the leading and trailing edges of the audio phrases that make up a sentence can be determined easily and under the same conditions, when they are combined and played back as a sentence, they can produce extremely natural sounds. It has the advantage of being able to create voice phrases that sound like

[Brief explanation of drawings]

FIG. 1 is a block diagram of a digital audio signal storage device according to an embodiment of the present invention. FIG. 2 is a memory map diagram of audio data stored in the audio memory. DESCRIPTION OF SYMBOLS 1...Analog-digital converter, 2...Audio memory, 3...Control unit, 4...Counter, 5...Silent pattern detection unit, _l1 , _l2 ...Signal line, _S1 ...Analog audio signal , _S2 ...Writing start signal, _S3 ...Writing stop signal, _S4 ...Silent frame instruction signal.

Claims (1)

[Claims] 1. An analog-digital converter that converts an input analog audio signal into a digital audio signal, an audio memory unit that stores the digital audio signal from the analog-digital converter in units of frames, and the audio memory. A digital voice signal storage device for a voice response device, which is connected to a readout output of the voice memory unit and outputs a detection output when the readout output is a silent pattern. The control means includes a means for temporarily reading out the digital audio signal written in the audio memory section and providing it to the audio pattern detection section, and an output of the silence pattern detection section. means for determining a frame in which a frame exceeds a certain percentage in one frame as a silent frame; A storage device for a digital audio signal, comprising means for editing audio phrases by deleting silent frames.