EP0610282A1

EP0610282A1 - Process for simultaneously transmitting audio signals from n-signal sources

Info

Publication number: EP0610282A1
Application number: EP92921746A
Authority: EP
Inventors: Karl-Heinz Brandenburg; Heinz GERHÄUSEN; Dieter Seitzer; Thomas Sporer
Original assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Current assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Priority date: 1991-10-31
Filing date: 1992-10-28
Publication date: 1994-08-17
Anticipated expiration: 2012-10-28
Also published as: RU2108001C1; DE59209456D1; KR100268517B1; JPH07504539A; DE4135977C2; CA2122577A1; NO941595D0; DK0610282T3; CA2122577C; AU2806992A; AU666339B2; FI942000A; WO1993009645A1; FI113936B; NO316098B1; DE4135977A1; ES2121868T3; FI942000A0; NO941595L; ATE169791T1

Abstract

A process is disclosed for simultaneously transmitting signals from N-signal sources over a corresponding number of transmission channels. The individual signals are subdivided into blocks and the blocks are converted into spectral coefficients by transformation or filtering, and the latter are then subjected to a data reduction process. The invention is characterized in that the blocks belonging to the individual signals are subdivided into section. The momentary sections of all signals are processed together, the admissible disturbance is determined for each section by using a perception-specific model and the momentarily required total transmission capacity is calculated. The allocation of the maximum available transmission capacity for each individual signal is calculated from the total available transmission capacity and the total momentarily required transmission capacity. Each individual signal is coded and transmitted with the thus determined capacity.

Description

Method for the simultaneous transmission of signals from N signal sources

Description

Technical field

The invention relates to a method for the simultaneous transmission of signals from N signal sources via a corresponding number of transmission channels.

State of the art

Methods are known in which the individual (time) signals are divided into blocks and the blocks are converted into spectral coefficients by a transformation or filtering, which in turn are subjected to a data reduction process or are coded accordingly for data reduction. For this purpose, reference is made, for example, to the overview article "Perceptual Audio Coding" by Jörg Houpert in studio technology or the article "Data Diet, Data Reduction in Digitized Audio Signals" by Stefanie Renner in Elrad, 1991. Reference is expressly made to these overview articles and PCT laid-open specification WO 88/01811 to explain all of the terms and process steps not described in more detail here.

In a number of cases, it is now necessary to transmit signals from several signal sources simultaneously over a corresponding number of transmission channels. The simplest example of this is the transmission of stereo signals over two transmission called channels.

When transmitting signals from N signal sources over a corresponding number of transmission channels, the problem of dimensioning the transmission channels now arises:

If each individual transmission channel is dimensioned in such a way that it can transmit the “maximum bit stream”, comparatively much transmission capacity remains “unused” on average.

Now it is known from digital telephone technology, when transmitting signals from a plurality of signal sources over a corresponding number of transmission channels, to design the transmission channels only for an "average demand" and to compensate for short-term increased demand on individual channels by assignment from other channels . The assignment is made exclusively via the signal statistics.

With regard to the state of the art, "a digital speech interpolation method with prediction-controlled word division" by Dr. H. Gerhausen (1980), "a digital speech interpolation method with current priority allocation", by R. Woitowitc (1977) or "a digital speech interpolation method with block priority allocation" by G.G. Klahnenbucher (1978) referenced.

According to the invention, it has now been recognized that the methods used in digital telephone technology to compensate for a fluctuating demand in the transmission of a large number of signals via a corresponding The number of transmission channels does not provide good results if the digital signals to be transmitted are previously subjected to a data reduction, for example using the so-called OCF method.

Presentation of the invention

The invention is based on the object of specifying a method for the simultaneous transmission of signals from N signal sources via a corresponding number of transmission channels, with which "data-reduced signals" are transmitted via transmission channels which are only dimensioned for an "average need", without perceptible signals , ie for example, audible losses in signal capacity can be transmitted.

An inventive solution to this problem is specified in claim 1. Developments of the invention are the subject of the subclaims.

The invention is based on the basic idea that, in order to compensate for the fluctuating demand for the simultaneous transmission of signals from N signal sources via a corresponding number of transmission channels, the allocation to the individual signals is not carried out according to statistical criteria, but already in the method step in which the signals for data reduction are encoded to compensate for the fluctuating demand by means of appropriate measures.

This basic idea according to the invention is explained in more detail below with reference to an exemplary embodiment under the drawing, in which: 1 shows a block diagram to explain the method according to the invention, and

2a and 2b the signal structure according to the invention.

In the method according to the invention, the individual signals are divided into blocks and the blocks are converted into spectral coefficients by means of transformation or filtering. To compensate for the fluctuating demand, the blocks belonging to the individual signals are divided into sections, and the respective current sections of all signals are processed together. This is shown graphically in FIG. 1 by corresponding "function blocks".

Using a perception-specific model, which can be a psycho-acoustic model for the transmission of audio signals, for example, the permitted interference is determined for each section and the requirement for the currently required total transmission capacity is calculated from this. The calculation of the total transmission capacity, ie the necessary number of bits, is carried out simultaneously for all blocks. The allocation of the maximum available transmission capacity for each individual signal is calculated from the total available transmission capacity and the currently required total transmission capacity. The "bit number" allocated for each individual signal is used to encode the individual signal and, accordingly, to transmit this individual signal. In the simplest case, the transmission capacity required is only balanced between the channels. In the development specified in claim 2, there is a reserve of transmission capacity, a so-called bit reservoir, from which, in the event that the total transmission capacity required exceeds the transmission capacity available on average, an allocation of transmission capacity follows.

This bit reservoir is always filled when the requested transmission capacity is smaller than the available transmission capacity (claim 3).

In any case, in order to prevent the bit reservoir from growing too large, in the event that the transmission capacity is very much smaller than the available transmission capacity, bits are forcibly allocated to the individual channels (Claim 4). This compulsory allocation is preferably carried out only to the channels or signal sources that have registered a requirement that is greater than an average requirement. A requirement that is substantially greater than the average requirement means that these signals are much more difficult to encode than conventional signals.

In any case, it is preferred, according to claim 9, if an overall block is formed from all separately coded signals of the signal sources, which consists of a fixed area, which contains information from which the separation of the signals can be determined, and of several areas of flexible length that picks up the encoded signals. This is shown schematically in Fig. 2a. A further saving in transmission capacity is obtained by recognizing the same input signals and transmitting them only once using a suitable transmission format (claim 6). This is shown schematically in Fig. 2b.

In any case, it is possible to exactly determine the transmission capacity currently required or only to estimate it (claims 7 and 8).

In addition, it is possible to carry out the method according to the invention largely in parallel. For this purpose, it is preferred if, according to claim 11, the coding of the individual signals already takes place during the calculation of the allocation of the transmission capacity for each signal.

A further preferred realization of the basic idea according to the invention is specified in claim 11:

If the required transmission capacity exceeds the available transmission capacity and no assignment can be made from the bit reservoir, it is possible to increase the value of the permitted interference for all signals in such a way that the required total transmission capacity corresponds to the available transmission capacity does not exceed (claim 11).

The following is a numerical example of a procedure for audio signals. It is expressly pointed out that the basic idea according to the invention is not restricted to audio signals; rather you can also use video signals or other signals subject to a perception-specific assessment are treated similarly.

Example of a possible procedure for audio signals:

Let y (t) be samples of the audio signal.

1) The audio signal y is broken down into samples (y (t) which are digitized in a manner known per se. The digitized samples are broken down into blocks of length 2n which, in the selected embodiment, overlap blocks with overlap n are:

x (k, b) = y (b * n + k) for k = 0..2n (b block number).

2) Each block of length n is transformed into spectral coefficients by a transformation, for example a Fast Fourier transformation or a cosine transformation:

x (j, b) = SUM (l = 0..2n; x (l, b) - f (1) -cos (pi • (21 + 1 + n) (2j + l) / (4n))) for j = 0..n with f (l) = sqrt (2) -sin (pi ♦ (1 + 0.5) / (2n))

3) Each of the blocks is broken down into sections and the energy density calculated for each section:

E (i, b) = (SUM (k = a (i) + l..a (i + l X (k, b) 2)) / (a (i + l) -a (i)) for i = l ... c, the coefficients a (i) being taken from Table 1 below. 4) For each section, the permitted disturbance is calculated using a suitable psycho-acoustic model, with reference to which reference is made to the literature. The masking between the tapes results from the earth-deficient disturbance

T (i, b) = MAX (k * = l ... i-1; E (k, b) -z (i-k))

the masking in the band:

s (i ₄ , b) = max (E (i, b) • e (i), T (i, b))

and the masking between the blocks:

ss (i, b) = max (s (i, b-l (/ 16, s (i, b))

the required number of bits is then calculated for each block.

5) Calculation of the necessary number of bits for the block:

a) for coding as with OCF (Huffman coding):

p = p0 + SUM (i = l..C; (a (i + l) -a (i) • (s (i, b) / ss (i, b)))

b) for PCM coding (SNR = 6dB / bit):

For each section, a scale factor and the number of bits per sample are transmitted as additional information p = pO + SUM (i * = l..c; (a (i + l)) • 10/6 • log (E ( i, b) / ss (i, b)))

In the following, the meaningful values for the individual quantities or constants are are given:

n = 512 c = 23 pO = 1200 for OCF (average number of bits per block) pO = 345 for PCM (scale factors: 10 bits / section,

Coding of the number of quantization stages: 5 bits / section

TABLE 1

i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 a (i) 0 4 8 12 16 20 24 28 32 36 40 46 52 60 70 82

i 17 18 19 20 21 22 23 24 a (i) 96 114 136 164 198 240 296 372

TABLE 2

e (i) le-1 le 2 le 3 le 4 le 5 le 6 le-7 le-8 le-9 a (l)

0 for i> 9

TABLE 3

i 1 2 3 4 5 6 7 e (i) 0.0004 0.0004 0.0004 0.0004 0.0004 0.0004 0.0004

i 8 9 10 11 12 13 14 15 e (i) 0.002 0.002 0.002 0.004 0.01 0.015 0.025 0.04

i 16 17 18 19 20 21 22 23 e (i) 0.06 0.06 0.06 0.08 0.08 0.11 0.14 0.18 I The bit numbers are then assigned to the individual signals. For this purpose, it is assumed that k (k) bits are required for coding the K input signals, while the number of bits available is Psoii.

psum = SUM (ρ) k))

Now a case distinction is necessary:

1) if psum * = psoll

Every signal gets the requested number of bits: z (k) = P (k)

2) If psum <psoll

Each signal receives more than the requested number of bits: z (k) - (psoll / psum) • p (k) e.g. K = 2, psoll = 1600, p (l) = 540, p (2) = 660 psum = 1200 z (l> = 1600/1200 • 540 = 720 (180 bit more) z (2) = 1600/1200 • 660 = 880 (220 bit more)

3) If psoll> psum

Each signal gets less than the requested

Number of bits: a) for OCF: z (k) = (psoll / psum). p (k) b) for PCM:

The minimum number of bits for each signal must not be undercut: z (k) = pO + ((psoll-K-pO)) • (p (k) -pθ) - e.g. B. K = 2, psoll = 1600, p0 = 500, p (l) = 600, p (2) = 1200 then Psum = 1800 z (l) = 500 + (1600-2- 500) / (1800-2- 500) • (600-500) = 575

(25 bit less) z (2) = 500 + (1600-2-500 (/ (1800-2-500). (1200-500) = 1025

(175 bit less)

The following case distinction is required to correct the permitted interference if p-bits are requested for each signal, but z-bits are allocated:

1) If the number of bits allocated is equal to the requested number: no correction necessary.

2) If more bits were allocated than requested: For OCF: no correction necessary.

For PCM:

The number of bits available for quantization in each section is increased by / z-p) / 512.

3) If fewer bits were allocated than requested: For OCF: ss (i, b) = s (i, b) + (z-p0) / (p-oO • (ss (i, b) -s (i, b)) for p> p0 ss (i, b) = s (i, b) for p <= p0

For PCM:

The number of bits available for the quantization in each section is increased by (zp) / 512. With PCM, rounding bits per ATW to an integer is necessary: For this purpose, all bits / ATW are first rounded down to the next lower integer and the resulting bit total is determined.

If bits are still available, a bit / ATW starting from the lowest bands is made available in a first pass until the available number of bits is reached.

Example:

104 bits are available

still to be distributed: 10 bits

+1 +1 result: 5 6 The invention has been described above with reference to exemplary embodiments. Within the general inventive concept, of course, the most varied variations are possible:

It is thus possible to use a fixed total block length, filling bits being used or forwarding to coders which have not yet ended. It is also possible to use a flexible block length, in which a maximum block length is specified and in addition a time averaging takes place.

Claims

-. ύ.-enanspucne

1. A method for the simultaneous transmission of signals from N signal sources via a corresponding number of transmission channels, in which the individual signals are divided into blocks and the blocks are converted into spectral coefficients by means of transformation or filtering, which are subjected to a data reduction process, characterized by following features:

the blocks belonging to the individual signals are divided into sections,

- the current sections of all signals are processed together,

using a perception-specific model, the permitted interference for each section is determined and a requirement for the currently required total transmission capacity is calculated,

- From the total available transmission capacity and the currently required total transmission capacity, the allocation of the maximum available transmission capacity is calculated for each individual signal and each individual signal is coded and transmitted with this capacity determined in this way.

2. The method according to claim 1, characterized in that a reserve of transmission capacity (bit reservoir) is available, from which, in the event that the total transmission capacity required exceeds the transmission capacity available on average, an allocation is made.

3. The method according to claim 2, _1 ^** ; _

characterized in that the bit reservoir is filled when the requested transmission capacity is less than the available transmission capacity.

4. The method according to claim 3, characterized in that - in order to prevent the bit reservoir from growing too large - in the event that the requested transmission capacity is very much smaller than the available transmission capacity, bits are forcibly allocated .

5. The method according to claim 4, characterized in that the compulsory allocation takes place only when there is a need which is greater than a mean requirement.

6. The method according to any one of claims 1 to 5, characterized in that the same input signals are known and are transmitted only once by a suitable transmission format.

7. The method according to any one of claims 1 to 6, characterized in that the determination of the currently required transmission capacity is carried out exactly.

8. The method according to any one of claims 1 to 7, characterized in that the determination of the currently required transmission capacity is only estimated.

9. The method according to any one of claims 1 to 8, characterized in that an overall block is formed from all separately coded signals of the signal sources, which consists of a fixed area containing information from which the separation of the individual signals

REPLACEMENT LEAF can be determined, and consists of several areas of flexible length.

10. The method according to any one of claims 1 to 9, characterized in that the coding of the individual signals is already carried out during the calculation of the allocation of the transmission capacity for each signal.

11. The method according to any one of claims 1 to 10, characterized in that in the event that the requested number of bits exceeds the total number of available bits, the allowed interference for all signal sources is increased, so that a reduced bit Request results.

REPLACEMENT LEAF