JP2001169238A - Nonlinear editing device, nonlinear editing method, recording medium, test method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a nonlinear editing device that can apply rendering to audio data in real time. SOLUTION: A video read means 32 reads video data by a prescribed display time from a video recording unit 4 and stores the data to an output video queue buffer 36. An audio read means 33 reads audio data of a prescribed size from an audio recording unit 5 and stores the data to an output audio queue buffer 31. An output means 30 applies rendering to audio data of a plurality of channels in the output audio queue buffer 31 to fuse the video data in the output video queue buffer 36 by a prescribed display time and the audio data corresponding to the video data by the prescribed display time and to output the resulting data to an external device.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a non-linear editing apparatus for editing video and audio recorded as digital data on an information recording medium used by a computer or the like.

[0002]

2. Description of the Related Art With the advance of LSI technology and the increase in capacity and high-speed input / output of a hard disk (hereinafter referred to as "HDD") as an information recording medium, the era in which moving images can be handled on a computer has been introduced. Non-linear editing devices that take advantage of the characteristics have appeared.

In the original non-linear editing apparatus, an analog video signal and an analog audio signal input from a source data input source device (hereinafter referred to as an “input source device”) are digitally converted for each field or frame by a unique method. After compression and recording on an HDD, the data is generally edited as necessary, converted into an analog video signal and an analog audio signal, and output to an output destination device (hereinafter referred to as “output destination device”). Was.

[0004] However, in recent years, the DV system, which is a digital video standard, has been established as a digital VTR standard for business use to consumer use, and an interface (hereinafter, referred to as an input source device or output destination device of a non-linear editing apparatus). (Referred to as "I / F") can now be input and output using DV format digital data. This gives D
In a non-linear editing device compatible with the V system,
It is not necessary to perform / D conversion and D / A conversion, and as a result, it has become possible to transfer data at a speed exceeding 1x speed without deteriorating video and audio.

A feature of the DV system is that one frame composed of ten tracks is used as a data unit, and video data and audio data are packaged. That is, ITI (Insert and Track Informa
A track is composed of four sectors, namely, a sector, an audio sector, a video sector, and a subcode sector. The video sector stores video data, and the audio sector stores audio data. Detail is,
"Introduction to DVD & DVC Basic Chapter 18" (by Kazuaki Mochiki)
(The first edition of October 20, 1996).

FIG. 15 shows a conventional nonlinear editing apparatus and its peripheral devices. This non-linear editing apparatus is constituted by using a computer 9, a CPU 1, a PCI bus 2 as a host bus, a DV HDD 4 and an audio HDD 5 connected to a SCSI adapter 3, a memory 6, a n-speed transfer, It is composed of a DV stream I / O adapter 8 for inputting and outputting a DV stream at n times speed to a corresponding VTR (n: an integer of 1 or more) and control software 7 for performing various controls on these devices. ing.
The control software 7 is software that operates on the multitask OS, and is a recording control unit that controls a process of recording a DV stream input from the DV stream I / O adapter 8 on the HDD (hereinafter, referred to as “recording process”). 71, a file management unit 72 that manages a DV stream recorded in the HDD, an edit control unit 74 that controls processing for editing video and audio (hereinafter, referred to as “editing processing”) and manages editing information, And a playback control unit 73 that controls a process of outputting the DV stream recorded in the DV stream to the DV stream I / O adapter 8 (hereinafter referred to as “DV output process”).

The display 10, keyboard 11, and mouse 12 connected to the computer 9 are:
The monitor 14 and the speaker 15 are used when instructing various processes, and the monitor 14 and the speaker 15 are used as a device for confirming video and audio when recording source data in the HDD or outputting the edited result to DV.

Hereinafter, the recording process will be described in more detail.

While confirming the image on the monitor 14, the mouse 12 records the DV stream to be edited on the HDD.
Alternatively, when the user gives an instruction using the keyboard 11, C
Under the control of PU1, the recording control unit 71 included in the control software 7 is activated, and the recording control unit 71 first
The DV stream input from the DV stream I / O adapter 8 in units of one frame is temporarily stored in the memory 6.

[0010] Next, the recording controller 71 controls the DV stream stored as described above (hereinafter referred to as "DV data").
The audio data is copied from the audio sector and subjected to deshuffling as described below, and recorded on the audio HDD 5 in units of one frame, while the DV data is also recorded in the HDD 4 for DV in units of one frame.

The above-mentioned deshuffling is a process of converting shuffled data into time-series data. In other words, since the audio data stored in the audio sector is shuffled,
The data is temporarily converted to a time-series PCM (pulse code modulation) format.

Such audio separation processing may be performed after the input from the DV stream I / O adapter 8 is completed. That is, first, the DV data on the memory 6 is recorded in the DV HDD 4, and when the input from the DV stream I / O adapter 8 is completed, the DV HD
The same effect can be obtained by reading out the DV data on D4 to the memory 6 again and performing the audio separation processing.

As described above, by separating and managing the audio data from the DV data, it is possible to freely exchange and process audio.

That is, FIG. 10 shows an example of an editing screen displayed on the display 10 when editing the data recorded as described above. In the scene 1, time codes 00: 00: 00: 07: 00 to 00:00:00 : The sound effect A is superimposed on the section until 15:00 (simultaneous playback), and the time code is from 00:00:00 to 07:00.
This shows a state where the sound effect B has been edited so as to be superimposed on the section up to 0:13:00. Note that "Scene 1 L" shown in FIG.
-Ch "and" Scene 1 R-ch "
This corresponds to left and right channel audio data recorded on the V HDD 4. The sound effects A and B (sound data of 3ch to 6ch) are sound data to be used as sound effects at the time of editing, and are usually prepared in advance in the HDD.

Here, on the editing screen having six audio tracks, up to six audio data can be superimposed. However, it can be stored in the audio sector of DV data of the number of channels equal to or less than the upper limit, such as 2 channels for 48 KHz 16 bit audio data (hereinafter, the channel is referred to as “ch”) and 4 ch for 32 KHz 12 bit audio data. There is an upper limit on the number of channels of audio data. Therefore, when editing exceeding the upper limit is performed, audio data for DV output (audio data to be stored in an audio sector when outputting to DV) is newly generated.

For example, when the editing shown in FIG. 10 is performed on audio data of 48 kHz and 16 bits, the editing control unit 74 included in the control software 7 performs a volume control process on the audio data of each channel, and The audio data of channels 1ch, 3ch and 5ch (Fig. 16, (la), (lb), (lc)) are mixed (added) to generate new left channel audio data (Fig. 16, (ld)). Generate and right ch
2ch / 4ch / 6ch (FIG. 16, (Ra), (Rb), (R
The audio data of c)) is mixed to generate new right-channel audio data ((Rd) in FIG. 16), and the audio data thus generated is stored in the HDD as audio rendering data of scene 1.

Here, the editing result is transferred to a VTR compatible with n-times speed transfer.
When the user instructs to DV-output (reproduce) the editing result to transfer the data to the tape 13 and record it on a tape, the reproduction control unit 73 included in the control software 7 under the control of the CPU 1
Is activated, and the reproduction control unit 73 outputs the existing audio data (that is, the time code 00: 00: 00: 00 to 0: 0).
Section until 0: 00: 07: 0 and time code 0
After the audio rendering data generated as described above is connected to the audio data after 0: 00: 15: 00 and combined with the DV data (that is, shuffled and stored in the audio sector of the DV data), FIG. As shown in A), DV output is performed to the DV stream I / O adapter 8 at n times speed.

As described above, in the nonlinear editing device,
By generating the audio rendering data in advance, even if the editing exceeds the above upper limit, the editing result can be output to the DV without any trouble.

It should be noted that the video rendering shown in FIG. 17 (flow of the reproduction process) is basically the same as the audio rendering, so that the description is omitted here.

[0020]

However, the above-described audio rendering takes a considerable amount of time to complete with a very large load. That is, the above-described conventional nonlinear editing apparatus has a problem that the DV output cannot be started until a certain time required for rendering has elapsed after the completion of the editing process. This problem becomes conspicuous as the editing section increases.

Further, the conventional nonlinear editing apparatus has a large capacity HDD in order to prevent a problem that the editing result cannot be output to DV because the HDD does not have enough free space for storing the audio rendering data. Had to be kept.

The present invention has been proposed based on the above-described conventional circumstances, and has as its object to provide a non-linear editing apparatus capable of performing audio rendering in real time.

[0023]

The present invention employs the following means to achieve the above object. That is, according to the present invention, as shown in FIG.
Data recorded on HDDs 4 (1) to 4 (4) for
Alternatively, it is assumed that a non-linear editing device (computer) 9 is capable of editing audio data recorded on the audio HDD 5 as an audio recording unit as necessary.

Here, when the user instructs to output the edited result as a DV, the reproduction control unit 73 included in the control software 7 executes the video reading means (DV reading task) 32 shown in FIG. A read task 33 and an output means (DV output task) 30 are generated.

The video reading means 32 reads video data for a predetermined display time from the video recording units 4 (1) to 4 (4), and outputs a video queue buffer (output D).
V queue buffer) 36.

The audio reading means 33 reads audio data of a predetermined size from the audio recording unit 5 and stores the audio data in the output audio queue buffer 31.

The output means 30 renders the audio data of a plurality of channels stored in the output audio queue buffer 31 in parallel with the operations of the video read means 32 and the audio read means 33. The video data for the predetermined display time stored in the output video queue buffer 36 and the audio data corresponding to the video data for the predetermined display time are combined and output to an external device (not shown).

According to such a procedure, audio rendering can be performed in real time.

[0029]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a diagram showing a non-linear editing apparatus to which the present invention is applied and its peripheral devices. The configuration of the non-linear editing apparatus will be described below, focusing only on differences from the above-described conventional apparatus. [Recording process] First, when the user instructs to record DV data, the recording control unit 71 transmits the recording DV buffer 21, the recording audio buffer 22, the DV management buffer 26, and the audio management buffer 27 shown in FIG. The DV input task 20 and the DV recording task 23 are secured in the memory 6.
And a voice recording task 24 are generated.

The processing performed by the DV input task 20 will be described below.

First, the DV input task 20 assigns a free area address of the recording DV buffer 21 composed of 32 areas of 120 Kbytes (corresponding to the size of one frame) to D.
While acquiring from the V management buffer 26, two addresses of the free area of the recording audio buffer 22 consisting of 32 64 Kbyte areas are acquired from the audio management buffer 27 (FIG. 3, step S1). The acquisition of the addresses of the two areas for the audio data is performed on 1ch (left c).
This is because two areas are required: an area for storing the audio data of h) and an area for storing the audio data of 2 ch (right ch). Note that one of the recording audio buffers 22
The reason why the size of the area is 64 Kbytes will be described later.

Next, the DV input task 20 issues a DV input command to the DV stream I / O adapter 8 by designating the address of the recording DV buffer 21 acquired as described above, and temporarily enters a pause state (FIG. 3. Step S2 → S
3).

The DV stream I / O adapter 8 receiving the DV input instruction records the DV data for one frame.
When stored in the V buffer 21, the DV input task 20
Reflects the state change of the recording DV buffer 21 in the DV management buffer 26. For example, when DV data is stored in the first area of the recording DV buffer 21,
The DV input task 20 stores the address of the first area of the recording DV buffer 21 and the size of the DV data stored in the first area of the recording DV buffer 21 in the first area of the DV management buffer 26. Yes (Management buffer 2
6.27 is a structure capable of storing a buffer address and a data size). Note that each area of the DV management buffer 26 corresponding to each area of the recording DV buffer 21 where no DV data is stored, or a DV corresponding to each area of the recording DV buffer 21 where the stored data is not valid data. In each area of the management buffer 26,
"0" is stored as the area address.

Next, the DV input task 20 executes the above DV
Copy audio data from audio sector of data,
After deshuffling and storing in the respective areas of the recording audio buffer 22 corresponding to the addresses obtained as described above, the DV recording task 23 is woken up (FIG. 3, steps S4 → S5 → S7).

Here, the area of one in the one recording audio buffer 22 is 64 Kbytes as described above, whereas
The audio data corresponding to a frame is about 3200 bytes in the case of 48 KHz 16 bits, and is 6 bytes in 20 frames.
Approximate to 4K bytes. Therefore, the DV input task 20
Wakes up the audio recording task 23 when the audio data for 20 frames has accumulated in one area of the recording audio buffer 22 (FIG. 3, step S5 → S6), and changes the size of the audio data for 20 frames to audio. It is stored in the management buffer 27.

The above process is repeated while the DV input continues (FIG. 3, step S8: No → S1). On the other hand, when the DV input is stopped by the user instructing the stop of the recording process (FIG. 3, step S8: Yes), the DV input task 20 stores the buffer address "-1" in the DV management buffer 26, and At this time, the size of the audio data on the recording audio buffer 22 and the address of the recording audio buffer 22 are stored in the corresponding area of the audio management buffer 27, and further stored in the next area of the audio management buffer 27. The address “−1” is stored, and the DV input processing ends.

Next, the processing performed by the DV recording task 23 will be described.

First, the DV recording task 23 executes the HD for the DV.
When the addresses of the empty areas D4 (1) to D4 (4) are acquired from the file management unit 72 and temporarily suspended (FIG. 4, step S20 → S21), and then woken up by the DV input task 20 (FIG. 3). , Step S7), DV
The buffer address and the data size are obtained from the first area of the management buffer 26 (FIG. 4, step S2).
2).

Here, when the buffer address is "0".
If it is not any of “−1”, the DV recording task 23
Is to record DV data specified by the buffer address and the data size in the free space.
After issuing a recording command to the HDD 4 for
Similar processing is repeated for data (DV data specified by the buffer address stored in the second and subsequent areas of the DV management buffer 26) (FIG. 4, steps S23 → S24 → S25 → S22).

On the other hand, if the buffer address is "0", it means that no DV data is stored in the recording DV buffer 21, or that the stored data is not valid data. Therefore, D in this case
The V recording task 23 is in a sleep state until it is woken up by the DV input task 20 (FIG. 4, step S23 → S
21).

When the buffer address is "-1", it means that the DV data to be recorded has been completed. Therefore, the DV recording task 23 in this case
Notifies the file management unit 72 of the total size of all the DV data recorded so far as the data length (registers the file), and ends the DV recording process (FIG. 4, step S).
24 → S26).

Here, the recording command is transmitted to the DV HDD 4
(1) → 4 (2) → 4 (3) → 4 (4) → 4 (1)
・ Issues for serialization in this order. As a result, the DV data input from the DV stream I / O adapter 8 is divided into blocks of one frame size,..., And the blocks are stored in the DV HDD 4 (1) and the blocks are stored in the DV HDD 4 (1). 2)
The blocks are serially recorded in the order of the DV HDD 4 (3) (see FIG. 11).

Next, the processing performed by the voice recording task 24 will be described.

First, the audio recording task 24 is executed for the audio HD.
The address of the vacant area of D5 is acquired from the file management unit 72 and temporarily becomes inactive (FIG. 5, step S30 → S).
31).

Next, the voice recording task 24 is woken up by the DV input task 20 (FIG. 3, step S6).
), And acquires the buffer address and the data size from the first area and the second area of the voice management buffer 27 (FIG. 5, step S32). The recording audio buffer 22 specified by the buffer address (that is, the buffer 22)
As described above, in the first area and the second area of the recording audio buffer 22, the audio data of 1ch and the audio data of 2ch are stored for 20 frames each.

Here, the buffer address is "0".
If it is not one of “−1”, the voice recording task 24
Issues a recording command of audio data of 1ch to the audio HDD 5, then issues a recording instruction of audio data of 2ch, and then issues subsequent audio data (in the third and subsequent areas of the audio management buffer 27). Similar processing is repeated for audio data specified by the stored buffer address and data size (FIG. 5, step S33).
→ S34 → S35 → S36 → S32). The above 2c
In the recording instruction of the audio data of h, an area adjacent to the area where the audio data of 1ch is recorded is specified (described later).

On the other hand, when the buffer address is "0", a pause state is set (FIG. 5, step S33 → S3).
1) If the buffer address is "-1", the file is registered and the DV recording process ends (FIG. 5, step S34 → S37). This point is the same as in the case of the DV recording process, and a detailed description will be omitted.

As a result, the audio data has a size of about 64 Kbytes in a continuous area (see FIG. 12) on the audio HDD 5 in the order of 1ch → 2ch → 1ch of the next frame → 2ch of the next frame. Will be recorded. [Reproduction Process] The editing process is the same as the above-mentioned conventional one, so that the description will be omitted, and the reproduction process will be described below.

First, when the user instructs to output DV, the reproduction control unit 73 sets the c of audio data that can be output to DV.
After obtaining the h number X from the edit control unit 74, the X number shown in FIG.
Output audio queue buffers 31 and one output DV
In addition to securing the queue buffer 36 in the memory 6, the DV read task 32, 1 to Xch audio read tasks 33 (1) to (X), and the DV output task 30 are generated. The “X” is an integer of 2 or more, and the determination method will be described later. FIG. 6 shows a state in which one output DV queue buffer 36 has areas corresponding to a plurality of frames.

Hereinafter, the processing performed by the DV read task 32 will be described.

First, the DV read task 32 sends the DV read list information (see FIG. 13) to the edit control unit 74.
(Step S61 in FIG. 7).

The read list information is for DV, 1c
h audio, 2 ch audio, etc., exist for each of the DV data and the audio data of each channel, and the “number of streams” indicates the total number of streams to be read as “stream I”.
"D" indicates stream ID information, "start frame" indicates position information of a frame to start reading, and "number of read frames" indicates the number of frames to be read from the start frame. That is, the stream ID, the start frame, and the number of read frames are configured as one, and the number of streams is arranged.

Here, the DV read task 32 firstly stores the recording location information of the head stream (ie, the address on the DV HDD 4 where the stream of the head stream ID in the read list information is recorded) in the file management unit 72. To get from. Next, an offset address value based on the start frame information is added to the recording location information to obtain a read start address.

Further, the DV read task 32 further executes the DV HDD 4 based on the read start address.
(1) -4 (4) DV for one frame for each
A data read command is issued to temporarily suspend the operation (FIG. 7, steps S62 → S63 → S64 → S65 → S
66 → S67). This read command is transmitted to the HDD 4 for DV.
As for (1) to (4), they are issued almost simultaneously as asynchronous instructions.

When the reading is completed, the DV reading task 32 copies the DV data read as described above one frame at a time (a total of four frames) and outputs the DV data.
It is stored in each area of the queue buffer 36 (FIG. 7, step S68). This output DV queue buffer 36
Since the software is FIFO controlled, the DV data stored by the DV read task 32 is
It is read by the V output task 30 (described later).

When the DV data cannot be copied because the output DV queue buffer 36 is full, the DV read task 32 is suspended until a space is available (FIG. 7, steps S69 → S70 → S68). ).

The above processing is repeated until reading of one stream is completed (FIG. 7, step S71 → S7).
2: Yes), the same processing is sequentially performed up to the stream of the last stream ID (FIG. 7, step S73).

Next, the processing performed by the audio reading task 33 will be described focusing on the differences from the processing performed by the DV reading task 32.

First, the k (k: an integer from 1 to X) ch audio read task 33 (k) acquires the read list information for kch audio from the editing control unit 74, and manages the recording location information of the first stream in a file. It is acquired from the unit 72 (FIG. 8, steps S50 → S51).

Next, the kch voice read task 33
(K) obtains a read start address based on the start frame information. Further, the kch voice read task 33
In (k), a command to read 20 frames of audio data is issued to the audio HDD 5 based on the read start address, and the HDD 1 temporarily enters a pause state (FIG. 8, steps S52 → S53). As described above, the size of the audio data read by this read command is about 64 Kbytes, that is, about the upper limit of the transfer capacity of the bus 2 that can be transferred in one transfer process.

When the reading is completed, the kch audio read task 33 (k) writes 20 frames of audio data into a predetermined area of the kch output audio queue buffer 31 (k) one frame at a time (FIG. 9). 8, step S54).

The output audio queue buffer 31
Is controlled by FIFO in software.
When the audio data cannot be copied because the output audio queue buffer 31 (k) is full, the kch audio read task 33 (k) is in a sleep state until there is a space (FIG. 8, steps S55 → S56). → S54).

The above processing is repeated until reading of one stream is completed (FIG. 8, step S57 → S5).
8: Yes), the same processing is sequentially performed up to the stream of the last stream ID (FIG. 8, step S59).

Next, the processing performed by the DV output task 30 will be described. However, in the following description, the audio data is 48 KHz 16 bits, and the channel of the audio data is
It is assumed that the number is 6 ch.

First, the DV output task 30 is for the DV (1
) And audio (six) edit script information (see FIG. 13) from the edit control unit 74, and suspends until the output DV queue buffer 36 and all the output audio queue buffers 31 become full. (FIG. 9, step S79)
→ S80).

The edit script information exists for each of the DV data and the audio data of each channel similarly to the read list information, and includes “stream number”, “stream ID”, “start frame”, and “read frame number”. . “Reproduction start time code” indicates a time code for starting DV output, and “volume (unnecessary for editing script information for DV)” indicates an output volume when the current volume is set to 100. This “volume” may include information indicating whether to perform fade-in and fade-out, and information indicating a fade section when performing fade-in or fade-out.

The reason why the DV output task 30 is set to the hibernate state as described above (FIG. 9, step S80) is when data storage in either one of the queue buffers 36 and 31 is delayed for some reason. Even if there is no problem, D
This is because the V output can be continued. That is,
If the queue buffer whose data storage is delayed before the queue buffers 36 and 31 run out recovers, the DV output can be continued without any trouble.

Here, the rendering means 30r included in the DV output task 30 performs the following rendering processing. That is, first, the audio data located in the area at the head of the audio output queue buffer 31 (1) of 1ch at the time code 00: 00: 00: 01 is read out, and based on the volume information included in the edit script information. The sound volume adjustment processing is executed, and is temporarily stored in the left channel mixing buffer 39. Next, the contents of the mixing buffer 39 are copied as they are to the left channel rendering buffer 34 (FIG. 9, steps S81 → S83 →).
S85).

Next, if there are three-channel audio data at time code 00: 00: 00: 00: 01, the DV output task 30 also performs volume adjustment processing on the three-channel audio data in the same manner, and performs left channel mixing. The data is temporarily stored in the buffer 39. Then, as described above, left channel
The data is mixed with the audio data of one channel copied to the rendering buffer 34, and the result is stored in the left channel rendering buffer 34. (FIG. 9, steps S86 → S87 → S88). Furthermore, 5ch
If there is any audio data, the sound volume adjustment processing and mixing are also performed on the audio data of 5 ch in the same manner as described above (FIG. 9, steps S89 → S90 → S91).

As a result, the left-channel audio rendering data for one frame is generated based on the audio data of 1ch, 3ch and 5ch.

Next, the DV output task 30 checks the right channel (2ch,
Based on the audio data of 4ch and 6ch), the right channel audio rendering data for one frame is generated in the same procedure as described above and stored in the right channel audio rendering buffer 35 (FIG. 9, steps S92 to S100).

Here, the shuffling means 30s included in the DV output task 30 shuffles the left and right channel audio rendering data stored in the audio rendering buffers 34 and 35 as described above, and then writes the
w is combined with the DV data in the time code code 00: 00: 00: 01 and output to the DV stream I / O adapter 8 together with the DV output instruction (FIG. 9, step S
102).

When the above processing is completed for all the frames (FIG. 9, step S102: Yes), the DV output processing ends.

Here, the number of channels of the audio data is six.
The case where the channel is a channel has been described as an example.
Up to 7ch and 9ch (left ch)
The same volume adjustment processing and mixing as described above are performed for 8ch and 10ch (right ch).

When the output DV queue buffer 36 and the output audio queue buffer 31 in which the DV output data are stored become empty, the read task 32
33 rises as described above (see step S70 in FIG. 7 and step S56 in FIG. 8). In this way, each reading process can be performed without depending on the DV output timing, so that the efficiency is high.

By the way, when reproducing audio data,
Normally, audio data of the left channel and audio data of the right channel are paired and played back simultaneously. The present invention pays attention to this point, and prepares a task 31 for audio reading for each channel, and also sets an area where audio data of the left channel 1ch and audio data of the right channel 2ch are continuous (see FIG. 12).
Was recorded. In this way, the seek operation of the HDD is suppressed as much as possible, and efficient reading can be performed.

Also, audio data is recorded in units of 64 KB.
The reading is performed because the size that can be transferred in one transfer process of the multitask OS adopted by the nonlinear editing apparatus is 64 Kbyt. If transfer is performed with a size smaller than the maximum size that can be transferred by the bus,
Transfer processing load increases, and the efficiency decreases. On the other hand, if a transfer instruction is issued by specifying a size larger than the maximum transferable size of the bus transfer, the number of issuances increases. For example, a transfer command specifying 96 Kbyt is divided into a 64 KB transfer command and a 32 KB transfer command by the internal processing of the OS. When the maximum size of the bus transfer is changed, it goes without saying that the recording / reading unit of the audio data is changed accordingly. [Sound Channel Number Determination Test] As described above, according to the present non-linear editing apparatus, audio rendering can be performed in real time. It is realized only under the condition that it is less than the number. That is, when the number of audio channels exceeds the predetermined number, audio data necessary for DV output is output from the audio HDD 5 to the output audio queue buffer 3.
Since reading to 1 is not in time, an error of a dropped frame (see FIG. 9, step S82) occurs.

Therefore, when constructing the system, the audio HDD
When the HDD 5 is replaced with another HDD, when the performance of the CPU changes, etc., the number of audio channels capable of real-time audio rendering at n times speed while performing volume adjustment processing, mixing, and shuffling is determined. A test (hereinafter referred to as a “voice channel number determination test”) is performed.

There are two major factors that cause delays in reading data from the HDD: frequent seeks and data being recorded on the innermost track (the innermost track of the HDD). No. That is, if data recorded in the discontinuous area is read at the same time, the head of the HDD moves alternately between the areas, and the reading is delayed. Also,
Since the HDD adopts the CAV (Constant Angular Velocity) method, the rotation speeds of the outermost and innermost circumferences are equal, but the information amount of the innermost circumference is about 2/3 of the information amount of the outermost circumference. Therefore, when the data to be read is recorded on the innermost circumference, the read performance is reduced to about 2/3 as compared with the case where the data is recorded on the outermost circumference.

As described above, the test for determining the number of audio channels includes a random access test in which data is recorded so that a seek frequently occurs, and a test in which all data are recorded in HD.
Two of the innermost circumference tests recorded at the innermost circumference of D are performed, and the method will be described below with reference to FIG.

FIG. 14A shows an example of a data recording mode in the 4ch random access test. Here, the 1ch and 2ch streams are recorded at the head address on the hard disk, and the 3ch and 4ch streams are recorded at the innermost circumference in the format shown in FIG. 12, respectively. Note that the frame length of each stream is preferably about 3000 frames.

The time code 00: 00: 00: 00: 0
A readout list corresponding to 1 to 3000 frame times is prepared, and edit script information is generated so that all the channels 1 to 4 need the volume adjustment processing.

FIG. 14B shows an example of the data recording mode in the 4ch innermost circumference test. Here, the 3ch and 4ch streams are recorded on the innermost track, and the 1ch and 2ch streams are recorded on the second track from the inside. Other conditions are the same as in the case of the 4ch random access test.

A test is made as to whether or not a frame drop occurs when DV is output under the above conditions. If a frame drop occurs in any of the 4ch random access test and the 4ch innermost circumference test, it is determined that the 4ch real-time rendering is not possible.

On the other hand, if no frame drop occurs in any case, real-time rendering of 4ch is possible, and then it is tested whether or not real-time rendering of 6ch is possible. The 6ch real-time rendering method is the same as the 4ch real-time rendering except for the addition of the 5ch and 6ch streams (see FIGS. 14C and 14D), and a description thereof will be omitted.

Thus, 4ch → 6ch → 8ch →.
··· Increase the number of audio channels by 2ch at a time, and end the test when a frame drop occurs. The maximum number of audio channels that have passed the test is the number of audio channels that can guarantee DV output while performing real-time rendering at n times speed. This number of audio channels is registered in the editing control unit 74. In this way, even if the number of audio channels changes, it is possible to flexibly cope with the need to upgrade the control software 7 (as described above, the reproduction control unit 73 sets the number of audio channels X to the edit control unit). 74).

The voice ch determined by this test
If it is desired to edit more than the number of audio data, it is needless to say that the audio rendering data should be generated in advance as in the related art.

[Test Results] The results of the tests performed as described above are as follows. The test conditions are as follows: the transfer capacity of the SCSI bus is 40 Mbyt / sec, the rotation speed of the hard disk 5 (4) is 7200 rpm, and the capacity is 4 Gbyt. The maximum transferable capacity of the OS is 64 Kbyt as described above. Further, it is assumed that n = 4.

(1) Change of read conditions First, the storage area of each of the left and right channels is set as one frame unit on the hard disk 5, and the storage areas of the left and right channels are not adjacent to each other and are sequentially in accordance with the instruction of the OS. Remember.

In this state, the transfer from the memory 6 to the I / O adapter 8 is performed for a maximum of 5 frames of data (up to 5 frames even if each queue buffer 31 (k) has a remaining area of 5 frames or more). In this case, the upper limit of the number of audio data channels that can be executed in real time in both the random read test and the innermost read test was four.

In the above description, the maximum read frame is 2
When set to 0, it was possible to read up to 6 ch in the random read test.
There was a possibility that the remaining capacity of (k) was reduced and underflow occurred. Further, in the innermost peripheral read test under this condition, the read limit was 4 ch. Further, the space of the maximum possible transfer capacity (64 Kbyt) is set in each queue buffer 3.
Even if each read task 33 (k) issues a read instruction at the stage when it is 1 (k), the expected result was not obtained.

(2) Changing the write condition Even if the read condition from the hard disk 5 was changed as described above, a satisfactory result could not be obtained.
The writing conditions were changed as follows. In other words, on the hard disk 5, the storage areas of the left and right channels are set to 64, which is the maximum transferable size for one transfer using the OS.
Kbyt, and the recording areas of the left and right channels were arranged adjacent to each other.

When the maximum read frame is set to 20 under these write conditions, the random read test is performed for 6 frames.
It is possible to read up to ch.
Although the result that the remaining capacity of (k) was sufficient and there was no possibility of occurrence of underflow was obtained, in the innermost circumference read test under this condition, the read limit was 4 ch. Next, under the above write conditions, the maximum possible transfer capacity (6
In the case of reading at 4 Kbyt), satisfactory results were obtained up to 6 ch in both the random access test and the innermost peripheral read test.

Further, the transfer capacity of the SCSI bus is set to 80 Mbyt.
/ sec, the rotation speed of the hard disk 5 (4) is 7200 rpm, and the capacity is 30 Gbyt. Since the storage density of the hard disk 5 increases, the amount of data that can be read in one rotation increases, and the read speed of the hard disk also increases. Inevitably will be higher. Therefore, it is estimated that real-time rendering of 12 channels or more is possible.

[0096]

As described above, according to the present invention, since the audio rendering can be performed in real time, the DV output can be started immediately after the editing process is completed, and the editing operation can be performed efficiently. Also, since the audio rendering data is generated on the memory, there is no need to provide an audio rendering data storage HDD.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a non-linear editing apparatus to which the present invention is applied.

FIG. 2 is an explanatory diagram of a recording process according to the present invention.

FIG. 3 is a flowchart showing a DV input process according to the present invention.

FIG. 4 is a flowchart showing a DV recording process according to the present invention.

FIG. 5 is a flowchart showing a voice recording process according to the present invention.

FIG. 6 is an explanatory diagram of a reproduction process according to the present invention.

FIG. 7 is a flowchart showing a DV read process according to the present invention.

FIG. 8 is a flowchart showing a voice reading process according to the present invention.

FIG. 9 is a flowchart showing a DV output process according to the present invention.

FIG. 10 is a diagram illustrating an example of an editing screen in the non-linear editing apparatus.

FIG. 11 is a diagram showing a recording format of DV data according to the present invention.

FIG. 12 is a diagram showing a recording format of audio data according to the present invention.

FIG. 13 is a configuration diagram of read list information and edit script information according to the present invention.

FIG. 14 is an explanatory diagram of a voice channel number determination test according to the present invention.

FIG. 15 is a configuration diagram showing a conventional nonlinear editing device.

FIG. 16 is a conceptual diagram of a conventional reproduction process.

FIG. 17 is a diagram showing a flow of a conventional reproduction process.

[Explanation of symbols]

 Reference Signs List 1 CPU 2 PCI bus 3 SCSI adapter 4 HDD for DV (video recording unit) 5 HDD for audio (recording unit for audio) 6 Memory 7 Control software 71 Recording control unit 72 File management unit 73 Reproduction control unit 74 Editing control unit 8 DV stream I / O adapter 9 computer 10 display 11 keyboard (instruction input means) 12 mouse (instruction input means) 13 VTR compatible with n-times speed transfer 14 monitor for video confirmation 15 speaker for audio confirmation 20 DV input task (input means) 21 recording DV buffer 22 recording audio buffer 23 DV recording task (video recording means) 24 audio recording task (audio recording means) 26 DV management buffer 27 audio management buffer 30 DV output task (output means) 31 output audio queue buffer 32 DV Reading Out task (video reading means) 33 voice reading tasks (audio read-out means) 34 Left ch audio rendering buffer 35 right ch audio rendering buffer 36 outputs the DV queue buffer

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G11B 27/02 K

Claims (16)

[Claims] 1. A non-linear editing apparatus capable of editing video data recorded in a video recording unit or audio data recorded in an audio recording unit as required, comprising: a video for a predetermined display time from the video recording unit; Video reading means for reading data and storing the data in an output video queue buffer; audio reading means for reading audio data of a predetermined size from the audio recording unit and storing the data in an output audio queue buffer; In parallel with the operation of the audio reading means, after rendering the audio data of a plurality of channels stored in the output audio queue buffer, the video data for a predetermined display time stored in the output video queue buffer Is combined with audio data corresponding to the video data for the predetermined display time. A non-linear editing apparatus, comprising: output means for outputting the output to an external device. 2. The non-linear editing apparatus according to claim 1, wherein said audio reading means reads audio data having a maximum block size of a bus transfer of an operating system from an audio recording unit. 3. The audio reading means alternately reads audio data of a left channel and audio data of a right channel from an adjacent area of the audio recording unit.
The nonlinear editing device according to 1. 4. The nonlinear editing apparatus according to claim 1, wherein said audio reading means is provided for each audio channel. 5. The non-linear editing apparatus according to claim 4, further comprising an instruction input means capable of inputting an instruction of the number of audio channels. 6. An input means for separating and storing digital data in a format in which the video data and the audio data are packaged in a recording video buffer and a recording audio buffer, and an operation of the input means. In parallel with the operation of the input means, video recording means for recording video data for a predetermined display time stored in the recording video buffer in the video recording unit. 2. The non-linear editing apparatus according to claim 1, further comprising: audio recording means for alternately recording left channel audio data and right channel audio data in a block size in an adjacent area of the audio recording unit. 7. A non-linear editing apparatus capable of editing video data recorded in a video recording unit or audio data recorded in an audio recording unit as required, wherein the video data obtained from the video recording unit is A video queue buffer for storing the audio data obtained from the audio recording unit, and / or an output audio queue for storing the rendering target audio data obtained from other resources in a frame unit; A buffer, rendering means for rendering audio data of a plurality of channels stored in the output audio queue buffer, video data in units of one frame stored in the output video queue buffer, Output means for combining audio data and outputting the combined data to an external device; Nonlinear editing apparatus characterized by comprising a. 8. The process of writing the output video cue buffer from the video recording unit to the output video cue buffer and the process of writing the audio data obtained from the audio recording unit to the output audio cue buffer. 8. The non-linear editing apparatus according to claim 7, wherein the non-linear editing apparatus is executed in parallel with the rendering processing by the rendering means and the output processing by the output means. 9. A non-linear editing method capable of editing video data recorded in a video recording unit or audio data recorded in an audio recording unit as required, wherein the video data for a predetermined display time is read from the video recording unit. A video reading process for reading data and storing the data in an output video queue buffer; an audio reading process for reading audio data of a predetermined size from the audio recording unit and storing the audio data in an output audio queue buffer; In parallel with the operation of the audio reading means, a process of executing rendering on audio data of a plurality of channels stored in the output audio queue buffer; and a video for a predetermined display time stored in the output video queue buffer. The data and the video for the predetermined display time after rendering Non-linear editing method for an output process of outputting to an external device combines the audio data corresponding to the data, comprising the. 10. The non-linear editing method according to claim 9, wherein said audio reading process reads audio data having a maximum block size of a bus transfer of an operating system from an audio recording unit. 11. The non-linear editing method according to claim 9, wherein said audio reading means alternately reads audio data of the left channel and audio data of the right channel from an adjacent area of the audio recording unit. 12. The digital data in a format in which the video data and the audio data are packaged is recorded in a video recording unit, and the audio data of the left channel and the right channel are stored in a bus transfer maximum block size of an operating system. 10. The audio data of a channel is alternately recorded in an adjacent area of an audio recording unit.
Non-linear editing method described in. 13. A recording medium on which a non-linear editing procedure capable of editing video data recorded on a video recording unit or audio data recorded on an audio recording unit as necessary is provided. A video read process for reading video data for the display time and storing the video data in the output video queue buffer; an audio read process for reading audio data of a predetermined size from the audio recording unit and storing the audio data in the output audio queue buffer; Processing for executing rendering on the audio data of a plurality of channels stored in the output audio queue buffer in parallel with the operations of the video read unit and the audio read unit; and a predetermined process stored in the output video queue buffer. The video data for the display time and the rendering Predetermined display time period recording medium and an output process of outputting to an external device combines the audio data corresponding to video data, the computer is recorded as readable program. 14. A test method for determining the number of audio channels of a non-linear editing device capable of outputting digital data to an external device in real time, wherein the efficiency of reading audio data from an audio recording unit is the worst. A test method characterized by performing. 15. The test method according to claim 14, wherein the test is performed in a state where audio data is recorded in an audio recording unit so that a seek frequently occurs. 16. The test method according to claim 14, wherein the test is performed under the condition that the audio data is recorded on the innermost circumference of the audio recording unit.