JPS5912187B2 - Audio editing output device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an audio editing and output device that outputs an audio signal that has been differentially compressed and encoded using an ADPCM (adaptive differential PCM) method or the like.

FIG. 1 shows a general configuration of an audio editing/output device that outputs audio signals subjected to differential compression encoding.

In FIG. 51, the editing control unit 1 receives an external audio output command S corresponding to audio output, and according to the content, sequentially reads and edits the contents of the corresponding audio file in the audio file memory 2 recorded as a digital signal. and sends it to the audio decoding circuit 3. The audio decoding circuit 3 restores the encoded data to the original analog audio signal V and outputs it in synchronization with the sample clock applied at the same time. In the audio file memory 2, files are created for each unit of audio information, such as a word. Audio file creation 1
The five generation device 4 is a means for creating data for the audio file memory 2 from the original audio, and its configuration is shown in FIG. The original audio signal is PCM encoded by the AD converter 5 and input to the audio processing section 6. The audio processing unit 6 cuts the original audio into audio information units, compresses the data by compression encoding such as ADPCM code, and writes the data into an audio file T with an appropriate label attached to each file.
The audio extracted by the audio processing unit 6 can be decoded by the DA converter 8 and monitored if necessary. The information of the .25 audio file T is transferred and stored in the audio file memory 2 of the audio output device.

FIG. 3 is a schematic diagram of conventional audio output edited and synthesized by such a system. In Figure 3, W3, W2
, W3 is the word string read out from the audio file 7.
In the expression, pauses P, . . . P2 are inserted between each word, and continuous audio output is obtained in the order of word strings Wl, W2, W3, . . . . Here, the invention of the present application will be explained using the ADPCM method as an example of compression encoding.
Of course, other compression encoding methods such as the 35ADM (adaptive delta modulation) method can also be considered in the same way.
In the ADPCM method, the quantization step size (hereinafter referred to as Δ) of a certain sample is determined by multiplying the quantization step size Δ value of the immediately preceding sample by a predetermined prediction coefficient corresponding to the ADPCM encoding code value. It is decided. Of course, upper and lower limits are set so that the quantization step size Δ value does not become extremely large or small, and the initial Δ at which ADPCM encoding starts is always a constant initial value ΔSTART (normally Δ (set to the lower limit value ΔMIN).
When the ADPCM method is applied to an audio editing/output device, it is necessary to initialize the sample quantization step size Δ for each audio information unit (word in this case) to be edited for reasons described later. In an ADPCM encoder or decoder, when the input information is continuous data indicating silence, the quantization step size Δ gradually decreases until ΔM
It will reach IN and be automatically initialized. Conventionally, in audio editing output devices, the quantization step size Δ
As described above, as a means of initializing the decoding circuit 3, the decoding circuit 3 is automatically initialized by inserting a pause as a fairly long silent interval between words, and each of the audio file memory 2 A method was used in which a specific code indicating a reset for initialization was inserted at the beginning of each word along with the ADPCM encoded data for each file. The former has the problem of degrading audio quality due to the insertion of unnatural pauses, and increasing memory capacity by storing data for pauses in the audio file, and the latter has the problem of inserting a specific code indicating a reset. Since the code is inserted at the beginning of each word, there are problems such as a decoder is required on the decoding side to decode the code, and the memory structure becomes complicated or inefficient. The present invention proposes an audio editing/output device that solves the problems peculiar to these compression encoding methods, including the ADPCM method, without affecting the audio quality or the audio file memory configuration, and will be described in detail below.

To briefly explain the necessity of initializing ADPCM encoded data for each word, the first reason is when ADPCM encodes an audio file and when reading and decoding an encoded audio file. The point is that the internal states of the ADPCM encoder and the audio decoder must be completely equivalent.

The second reason is a problem when an error occurs in the ADPCM encoded data during the process until the ADPCM encoded data is input to the decoding circuit. For example, a series of ADs for a word
When an error occurs in a sample of PCM encoded data, the decoded output of all subsequent samples is affected, and the desired waveform is no longer reproduced. This is clear because the ADPCM encoded data is a differential signal and the decoding quantization step size Δ is successively changed depending on the history of the input code data series. Because of the need to stay within one word. For this reason, it is necessary to initialize the speech decoding circuit 3 for each word to be edited and output. FIG. 4a is a time chart of the edited audio data output of the audio editing output device according to the present invention and the initialization flag of the decoding circuit.

In other words, the words Wl, W2, W3, etc. are output consecutively as shown in the figure, there is no particular pause between the words, and the initialization flag is set at the same timing as the beginning of each word. Output. Of course, pauses necessary for voice generation may be inserted. FIG. 4b is an enlarged view of a part of FIG. 4a, which is a detailed time chart of the initialization flag, and shows the beginning of a certain word Wh. The sample clock indicates the timing at which ADPCM encoded data is sent, and a value of around 8 KHz is normally used. The initialization flag is sent at the first sample of each word, and even if ADPCM encoded data is present at this time, it is effectively invalid. FIG. 5 is a diagram showing a method of creating ADPCM encoded data in the audio file creation device 4.

The audio file creation device extracts words from the original audio in the form of a PCM signal, converts the words into a file, and finally converts the words into ADPCM encoded data. In a method that first performs extraction from ADPCM-encoded data, the internal state of the ADPCM-encoded data (for example, what value the quantization step size Δ is) must be considered during extraction. This is because it is difficult to extract and the operation becomes complicated. FIG. 5a shows an example waveform of a PCM audio file. The numbers indicate sample numbers, where 1 indicates the first sample cut out in the cutout operation and n indicates the last sample, so the data word length is n. For example, if the sample timing is 8 KHz, the sample interval corresponds to a vocalization time of 125 μs. ADPCM encoding is performed here, but prior to that, the PCM waveform is partially modified. There are two methods for this; the first method is to reset the PCM code value of the first sample (sample number 1) to 0" in FIG. 5b. In Figure 5c, the PCM audio file is shifted by one sample, and 10'' is inserted into the first sample (sample number 1). The data word length n remains unchanged, in which case the last nth sample data is lost as a result of being shifted. ADPCM encoding processing is applied to the PCM data manipulated in this manner from a state in which the quantization step size Δ is initialized. FIG. 5d shows the difference data between samples in the case of FIG. 5b. The first sample starts from a state where the audio is initialized (that is, when there is no sound), so the difference is 0. The second sample becomes the difference information from '0'.

For example, the difference D between the 6th and 7th samples in Figure 5b
corresponds to the seventh differential sample value D in FIG. 5d. Above, A is added to the PCM data shown in Figure 5b.
The case where DPCM encoding processing is applied is shown, and Fig. 5c
When the PCM data shown in FIG. 5 is subjected to ADPCM encoding processing, the result is almost the same, and the result is a form obtained by shifting the series of inter-sample difference data shown in FIG. 5d by 1 bit.

Using the above method, ADPCM encoding is performed for each word to create a final audio file. As shown in Figure 4, the decoding circuit is initialized by receiving an initialization signal at the same time as the first sample of each word is received, and by using such a data format, initialization can be performed without any problems. Since there is no pause between each word, the sound quality is not affected at all even at word boundaries, making it possible to realize ADPCM sound editing.

Next, a circuit for editing and outputting an audio file and creating an ADPCM encoded data initialization signal will be described.
FIG. 6 is a diagram showing the main parts of the audio editing output device of the present invention.

In FIG. 6, the editing control section 1 is connected to the input interface 9.
, decoder 10, audio file address table 11,
Audio file readout control unit 12, 1j single leg circuit 13,
It consists of a buffer memory 14 and an initialization flag memory 5. Audio output command data S from the outside is taken in by an input interface 9 and manually inputted to a decoder 10. Each word stored in the audio file memory 2 is assigned a unique word name. (For example, in Figure 6, W
(1, W2, W3, etc.) The decoder 10 decodes the voice output command data S and converts it into a corresponding word example. That is, the decoder 10 has a substantial editing function. The audio file address table 11 stores the storage address of the audio file memory 2 for each word name as its start address and end address, and stores the start address and end address for each word of the word string report output from the decoder. The audio file readout control unit 12
and stores it in an address register (not shown) in the audio file readout control section 12. The audio file reading control unit 12 sequentially reads the ADPCM encoded data stored in the audio file memory from the first address set in the address register to the final address for each word, detects the final address, and reads the word. Finish reading the word and move on to reading the next word. These entire operations are controlled by a control circuit 13. In the audio file memory 2, each word is stored in address order, and each word is divided into a plurality of blocks according to the reading unit. Each block data is read out all at once and stored in the buffer memory 4, and then read out one sample at a time according to the sampling period and input to the audio decoding circuit 3. Batsufuamemo l river 4
may be a toggle type in which writing and reading are performed alternately. An initialization flag memory 15 is provided together with a buffer memory 14. An initialization flag is set in the initialization flag memory 15 by a command from the audio file readout control unit 12 when data is written into the buffer memory 14, but this is limited to the same timing as when writing the first block of each word. It is not set when writing to other programs. Whether it is the first block or not can be easily determined from the first address information read from the audio file address table 11 by the audio file read control section. When a program is read from the buffer memory 14, the information in the initialization flag memory 15 is transferred to the audio decoding circuit 3 together with the first sample, and the initialization flag is then reset. As described above, according to the audio editing output device of the present invention, audio decoding and initialization at the beginning of each word can be performed without any problem in the ADPCM system. Also, in the explanation, ADPC
Although the M method is described as an example, it can be similarly applied to other compression encoding methods. The present invention provides a means for adding basic audio editing functions to an audio output device using a compression encoding method and improving audio quality and audio file memory efficiency. The effect is great when applied to various audio output devices.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an audio output device, Fig. 2 is a diagram showing means for creating audio file memory data from original audio, Fig. 3 is a schematic diagram of audio output, and Fig. 4 is an audio editing method of the present invention. A diagram showing a time chart of edited audio data and an initialization signal of the output device, FIG. 5 is a diagram showing a method of creating ADPCM encoded data, and FIG. 6 is a diagram showing the editing control unit and audio file of the audio editing output device of the present invention. FIG. 3 is a diagram showing main parts of a memory. 1...Editing control unit, 2...Audio file memory, 3...Audio decoding circuit, 4...
...Audio file creation device, 5...AD converter, 6...Audio processing unit, 7...Audio file, 8...DA converter, 9...Input interface, 10...Decoder, 11.
...Audio file address table, 12...
...Audio file reading control section, 13...
Control circuit, 14... Buffer memory, 15...
...Initialization flag memory.

Claims (1)

[Claims] 1.Equipped with an audio file memory in which each audio signal serving as a unit of editing is differentially compressed and encoded as each audio file, and in which the first differentially compressed encoded sample is initialized and stored, and read from this audio file memory. An audio editing/output device that edits an audio file and then decodes it in a decoding circuit to output an audio signal includes a buffer memory that temporarily stores the differential compression code read from the audio file memory, and a buffer memory that temporarily stores the differential compression code read from the audio file memory; an initialization flag memory that stores an initialization flag for initializing the internal state of the circuit; and an initialization flag memory that stores an initialization flag for initializing the internal state of the circuit; and means for setting an initialization flag and providing the initialization flag to the decoding circuit in synchronization with reading the first differential compression encoded sample of the audio file from the buffer memory. Output device.