WO2009124779A2 - Method of optimizing transmission of a stream of audio frames with variable bit rate in a transport stream for transmission on a network - Google Patents

Abstract

The invention relates to the optimization of the transmission of a stream of audio frames with variable bit rate in a transport stream, comprising: - a reception of at least one audio frame (62) of variable bit rate; - a generation of a time reference (63) for each audio frame belonging to a set of frames received, the time reference being associated with a moment of reception of the frame; - an encapsulation of at least one frame in an elementary stream packet (641) with the time reference generated, the packet having a size determined as a function of characteristics of the stream of audio frames and as a function of a payload of transport packets of the transport stream; - an encapsulation of at least part of the elementary stream packet in at least one transport packet (642); and - a transmission (65) of at least a part of a transport packet.

Description

 A method of optimizing transmission of a variable rate audio frame stream in a transport stream for transmission over a network.

1. Field of the invention

The present invention relates to the field of variable bit rate digital audio stream transmission to one or more receivers via a packet transport network. More specifically, the invention relates to reducing the bit rate of a variable bit rate (VBR) stream for sending it in a transport stream for transmission over a network.

2. Technological background.

According to the state of the art, an encapsulator of audio frames allows the packaging of these frames in view of its adaptation for sending packets on a network. The tasks of such an encapsulator include the packaging of an Audio Frame Stream (or "Audio Frame Stream") into Packetized Elementary Stream ("Packetized Elementary Stream") packets. The PES audio stream (s) are then put into transport packets, for example according to the MPEG2-TS format (Moving Pictures Expert Group, Transport Stream, the MPEG2-TS format is described in document ISO / IEC 13818-1). , allowing transmission of the audio frame stream over a packet transport network.

The audio frames processed by such an encapsulator may be of constant bit rate or of variable bit rate. Generally, the management of this type of network imposes a given flow limited flow. Bandwidth allocated to a stream transported by a packet transport network (for example DVB-H (Digital Video Broadcasting - Handheld) or portable devices "in French) or 3GPP (" Third Generation Partnership Project ") is limited for technical, economic or regulatory reasons.

The process of encapsulating a flow of variable rate audio frames into a transport packet stream causes a delay. In the context of the transport of audio frames by a packet transport network for mobile telephony, a short conversion period is advantageous because the delay between sending and receiving such a stream will be reduced, which increases the comfort of use of such a network.

According to the state of the art, the conversion of a stream of variable rate audio frames into a transport packet stream is done by adding stuffing bits (or stuffing bits) when packaging the packet. or audio PES packets in a transport stream packet, especially to ensure that a new audio PES packet starts at the beginning of a transport stream packet, which simplifies the encoding of such a stream and its decoding, but this does not optimize the bandwidth of the stream of audio frames thus converted.

According to the state of the art, when converting a stream of variable rate audio frames into a stream of transport packets, and especially in the packaging of variable rate audio frames, and therefore variable length, the the number of audio frames per packet of elementary stream is fixed, again to facilitate encoding and decoding, but this leads to an encoding delay, because before being able to send a packet of elementary stream thus constructed, the header of this elementary stream packet must be built. The header includes the size of the elementary stream packet, which is variable according to the state of the art because it depends on the size of the received audio frames. This process leads to a temporary storage of a number of audio frames before sending in one or more transport stream packets. This temporary storage ("bufferization" in English) results in an encoding delay. Thus, the prior art does not optimize this delay. As previously mentioned, the maximum throughput of a transport packet stream is in practice subject to maximum throughput constraint. In this context, adding stuffing data reduces the available bandwidth for the audio data, which has an impact on the quality of the transmitted audio data. Also, in the case where an audio PES stream is put into a transport packet stream together with a video stream, adding stuffing data to the audio PES packets reduces either the maximum available bit rate for the audio stream, or the video stream, on both.

Also, the state of the art has the disadvantage of a non-optimized conversion of a stream of variable rate audio frames in a stream of transport packets. 3. Summary of the invention.

The invention aims to overcome at least one of these disadvantages of the prior art.

More particularly, the invention aims to optimize the transmission of variable rate audio frames in a transport packet stream.

One of the optimizations of the transmission obtained by the invention lies in the reduction of the bandwidth taken for the transmission of a transport packet stream constructed according to the invention, in particular by an optimal filling of the elementary flow packets. and transport stream packets. By optimally filling the elementary stream packets and the transport stream packets, the method for optimizing the transmission of variable rate audio frames makes it possible, in particular, not to add stuffing data as well in the packetization step. elementary stream of variable rate audio frames only in the elementary stream packet transporting packetization step. To obtain this optimization, the transmission method of the invention comprises a step of determining a size of a packet of elementary streams, optimized at the same time to the characteristics of the stream of received audio frames and to the characteristics of the transport stream.

Another optimization of the transmission obtained by the invention lies in the reduction of the transmission delay of a variable rate audio frame received at the input of a transmitter packaging these frames in a signal stream. elementary packets and the packaging of elementary packet streams in a transport stream. This optimization is obtained in particular by setting the size of the elementary stream packets according to the method of the invention, and by calculating the time reference of these packets according to the invention, which then makes it possible to know very quickly the parameters which must be put in the header of a packet of elementary stream, and thus be able to transmit audio frames received on the fly, that is to say that the transmission method according to the invention requires only temporary storage of a single audio frame before it is transmitted in a transport packet stream, which reduces the transmission delay of the audio frames.

For this purpose, the invention proposes a method for optimizing the transmission of a stream of variable rate audio frames in a stream of transport for transmission over a packet transport network. In order to optimize the transmission, the method includes the following steps:

receiving at least one audio frame (62) of variable bit rate;

generating a time reference (63) for each audio frame belonging to a set of received variable rate audio frames, the time reference being associated with a reception timing of the audio frame;

encapsulating at least one variable rate audio frame in an elementary audio stream packet (641) with the generated timing reference, the audio elementary stream packet having a size determined according to characteristics of said audio frame rate stream variable and depending on a transport packet payload of the transport stream; encapsulating at least a portion of the audio elementary stream packet in at least one transport packet (642); and

- Transmission (65) of at least a portion of a transport packet. Advantageously, the transport packet constructed according to the method does not include a stuffing bit added to the audio component present in the transport stream packet,

- neither in audio elementary stream packets,

nor added to the audio elementary stream packet to encapsulation in at least one transport packet.

A transport packet stream constructed according to the transmission optimization method of the invention, comprising only variable rate audio frames, will therefore also be variable bit rate. Then, the transport stream according to the invention can be converted into a constant bit rate transport stream if, for example, the transport stream also comprises a component other than the audio component, for example a video component and / or a data component. which stuffing data is added to make the total transport stream (audio + other components) at a constant rate, a technique included in the state of the art. Advantageously, the method comprises a determination of the size of the elementary stream packet according to at least the following criteria: a first minimum size of the out-of-box elementary stream packet is equal to the maximum size of the audio frames of the audio frame stream,

a second minimum size of the elementary stream packet is greater than or equal to the size of a forward transport packet of the transport stream,

the third maximum size of the elementary stream packet is an integer times the size of a forward transport packet of the transport stream, and the size of the elementary stream packet is a value that is equal to or greater than the larger of the first and second minimum sizes and which is equal to or less than at least the third maximum size.

According to a specific embodiment, the method comprises a step of determining a fourth maximum size of the elementary stream packet as a function of a maximum decoding delay of the transport stream, and that the size of the elementary stream packet is d a value that is equal to or less than the smallest of the third and fourth maximum sizes.

According to a specific embodiment, the method comprises a step of prior analysis of the characteristics of the flow of variable rate audio frames and that said size of said packet of elementary audio stream is determined according to the result of this analysis.

According to a specific embodiment, the value for the size of the audio elementary stream packet is determined from characteristics of the audio frame stream provided to the transmission optimization method.

According to a specific embodiment, the encapsulation step of at least one audio frame, and the encapsulation step of at least a part of the audio elementary stream packet are combined in a single encapsulation step. variable rate audio frames in transport packets.

According to a specific embodiment, the encapsulation step of at least one audio frame, and the step of encapsulating at least a portion of the audio elementary stream packet are distinct.

According to a specific embodiment, the step of generating a time reference comprises the following steps:

when the payload portion of an audio elementary stream packet begins with the start of an audio frame, the time reference for the header of an audio elementary stream packet is equal to the time reference corresponding to the time of reception of the audio frame; and

- when the payload portion of the audio elementary stream packet does not start with the start of an audio frame, the time reference for the header of a current audio elementary stream packet is equal to the sum of a reference of corresponding time at the time of reception of the beginning of the audio frame and the duration of reception of the audio frame. The time reference of the header of an elementary stream packet makes it possible, on reception, to clock the decoding of the elementary stream. To allow such a timing, the time reference is included in a determined manner. According to for example the ISO / IEC 13818-1 standard, the time reference associated with the time of reception of the first complete audio frame included in the elementary stream packet.

4. List of figures.

The invention will be better understood, and other features and advantages will appear on reading the description which follows, the description referring to the appended drawings among which:

- Figure 1 shows a schematic block diagram of an infrastructure implementing the invention in two different types of equipment;

- Figure 2 shows an apparatus for implementing the invention according to a particular embodiment of the invention, such an apparatus belonging to the infrastructure of Figure 1;

FIG. 3 represents an exemplary format of audio frames accepted at the input of the apparatus of FIG. 2;

FIG. 4 represents an exemplary output data format provided by the apparatus of FIG. 2; FIG. 5 diagrammatically illustrates the different levels of encapsulation of the data processed by the apparatus of FIG. 2;

FIGS. 6 to 8 illustrate a method for optimizing the transmission of variable rate audio frames in a transport stream for transmission, according to the invention and implemented in the apparatus of FIG. 2; FIG. 9 illustrates a particular embodiment of a method of receiving a transport stream constructed according to the transmission method.

5. Detailed description of the invention.

FIG. 1 presents a schematic block diagram of an infrastructure 1 in two different types of equipment (encapsulator 104 and re-multiplexer 110) embodying the invention.

The infrastructure 1 comprises: an audio source 100;

an encoder 102 connected to the source 100 via a connection 101;

the encapsulator 104, connected to the encoder 102 by a connection 103;

a first receiver 108 connected to the encapsulator 104 by a first network 106 and connections 105 and 107;

a remultiplexer 110, connected to the encapsulator 104 by the network 106 and connections 105 and 109; and

a second receiver 114 connected to the remultiplexer 110 by a second network 112 and connections 111 and 113. The audio source 100 supplies audio data to the encoder 102.

The encoder 102 provides a variable rate audio frame stream to the transport packet encapsulator 104. The encapsulator 104 provides a transport packet stream, including a variable rate audio frame stream, to the first receiver 108 and The first receiver 108 is adapted to decode the stream of audio frames included in the transport stream. The remultiplexer 110 (sometimes also called rémux, re-encoder, or transcoder) is adapted to receive a stream of transport packets, multiplexed with other streams or not, to extract therefrom a stream of variable rate audio frames, and re-encode this stream of audio frames at another rate and / or according to another standard, and possibly re-multiplex with other streams. For example, the remultiplexer 110 re-encodes a stream of audio frames encoded by the encoder 102 to the AAC audio encoding format (according to ISO / IEC 14496-3: 2005) in audio encoding format AC3 (according to the ETSI standard TS 102 366). The re-multiplexer 110 then re-encapsulates the audio frame stream resulting from this conversion, to create a new transport packet stream comprising a variable rate audio frame stream, which is transmitted to the second receiver 114 via the second network 112. The second receiver 114 is adapted to decode the stream of audio frames included in the transport stream from the remultiplexer 110.

According to one variant, the remultiplexer receives a stream of multiplexed transport packets with other streams of audio or video type, extracts a stream of variable rate audio frames therefrom, re-encodes it according to another bit rate and / or according to another encoding standard, encapsulates it in a stream of transport packets and transmits it.

According to one variant, the remultiplexer receives a stream of multiplexed transport packets with other streams of audio or video type, and extracts a stream of variable rate audio frames therefrom, remultiplexing it with at least one other stream of audio or video type , encapsulates it in a stream of transport packets, and transmits it.

Alternatively, the remultiplexer receives a stream of transport packets, extracts a stream of variable rate audio frames, re-encodes it with another rate, encapsulates it in a stream of transport packets, and transmits it.

According to variants, the networks 106 and 112 comprise several audio encoders (for example two, three, four or more) and / or several encapsulators, several rémultiplexers, as well as other equipment useful for proper operation.

In order to simplify the drawing, the networks 106 and 112 are represented with a single receiver. According to variants, the networks 106 and

112 include a plurality of receivers (eg, two, three, ten, ...) adapted to receive and decode an audio packet stream including a variable rate audio frame stream.

According to variants of the invention, the transmission optimization method is implemented in the encapsulator 104, the re-multiplexer 110, or any other type of apparatus receiving a stream of variable rate audio frames, encapsulating these frames in transport packets (directly or via encapsulation in a stream of elementary packets), which are transmitted to one or more recipients.

Figure 2 schematically illustrates an apparatus 2 implementing the invention according to a particular embodiment. The apparatus 2 is for example the encapsulator 104 or the remultiplexer 110.

The apparatus 2 comprises, interconnected by an address and data bus 23: a microprocessor 21 (or CPU, from the English "Central Processing Unit");

a non-volatile memory of ROM type (of the English "Read OnIy Memory") 20; - a random access memory or RAM (from the English "Random Access

Memory ") 26;

an interface 22 allowing the reception of audio frames at the input;

an interface 24 for sending a stream of transport packets on a packet transport network; and

an internal clock 25 for associating a time reference with a received audio frame.

It is observed that the word "register" used in the description of the memories 20 and 26 designates in each of the memories mentioned, as well a memory area of low capacity (some binary data) that a large memory area (to store an entire program or all or part of the audio data received).

The ROM 20 includes in particular:

- a "prog" program 201; a multicast sending address (or "multicast" in English) 202;

a size value of the transport packets 203;

a size value of the header of the transport packets 204; and a size value of the header of the elementary stream packets

205.

The algorithms implementing the steps of the method described below are stored in the ROM 20 associated with the device 2 implementing these steps. At power up, the microprocessor 21 loads and executes the instructions of these algorithms.

RAM 26 includes in particular:

in a register 261, the operating program of the microprocessor 21 which is charged when the device 2 is powered up; a buffer zone for storing a received audio frame in a register 262; a time reference associated with the moment of receiving an audio frame, in a register 263;

a register 264 for preparing and storing a transport packet; a register 265 allowing the storage of the determined size of the elementary stream packets; and

a data zone 266 allowing the temporary storage of data necessary for the proper functioning of the apparatus 2. Other structures than that described with reference to FIG. 2 are compatible with the invention. In particular, according to variants, the invention is implemented in a purely hardware ("hardware") embodiment, for example in the form of a dedicated component (for example in an ASIC or FPGA or VLSI) (respectively " Application Specifies "Integrated Circuit" in English, meaning "Integrated Circuit for a Specific Application", "Field-Programmable Gate Array" in English, meaning "In-Situ Programmable Gate Network", "Very Large Scale Integration" in English , meaning "very large-scale integration")) or of several electronic components integrated in a device or in the form of a mixture of hardware elements and software elements ("software" in English).

FIG. 3 represents an exemplary format of audio frames 30, 31 and 32 successively received at the input of the apparatus 2. The example of FIG. 3 is presented as a simple illustration; the device 2 can receive other audio frames of different lengths and different formats.

The audio frames 30, 31 and 32 accepted at the input of the apparatus 2 are formatted according to the audio MPEG standard, and in particular MPEG 1 and 2 layer I, II and III (according to ISO / IEC 11172-3 references. ISO / IEC 13818).

The frame 30 (respectively 31, 32) comprises: a header (or "header" in English) 301 (respectively 311, 321), comprising fields of header 3011 to 3014 (respectively 3111 to 3114, 3211 to 3214). ); and

a field 302 (respectively 312, 322) of useful audio data (or "payload" in English) whose size is variable from one frame to another.

The letters in the header 301 (respectively 311, 321) represent bit fields; their meaning depends on the associated standard. According to MPEG audio standard mentioned above, the header 301 (respectively 311, 321) comprises four bytes 3011 to 3014 (respectively 3111 to 3114, 3211 to 3214). Bytes 3011 and 3012 (respectively 3111 and 3112, 3211 and 3212 include, for example, an 11-bit synchronization word (or "syncword" in English) All the bits of the synchronization word are set to "1", This makes it possible to recognize the start of a frame in a stream of audio frames.By 3012 (resp 3112, 3212) also includes two bits indicating the version of the MPEG standard, two bits indicating the value of the layer. of the MPEG standard, then a single bit indicating whether the header is followed by a value of sixteen bits of cyclic redundancy check (or "cyclic redundancy check" or "crc" in English) byte 3013 (or 3113). , 3213) includes an index value in a table according to the audio MPEG standard, representing the rate of coding of the frame ("bit rate" in English)

Byte 3013 (or 3113, 3213) also includes two bits for sampling rate (or "sampling rate" in English) for the frame, followed by two indicators of presence of an extension bit (" padding bit "in English) and a meaningless bit according to the MPEG audio standard. Byte 3014 (or 3114, 3214) includes an indicator for stereo or mono coding, followed by other bits indicating whether it is a copy, whether the frame content is subject to copyrights etc.

As mentioned above, the size of an audio frame is variable; the length of the frame can be calculated and fixed from information present in the header of the audio frame. FIG. 4 represents an exemplary output data format provided by the apparatus 2 and more particularly a transport packet comprising successively:

a transport packet header 40 and a payload comprising a packet of elementary streams; a packet of elementary streams with a header 41; and

a payload portion comprising three audio frames 30, 31 and 32.

According to the MPEG2-transport standard (ISO / IEC 13818-1), transport packets have a fixed length of 188 bytes, including a four-byte header. The header 40 of the transport packet is composed as follows.

First, a synchronization word 401, having a fixed value of 47 hexadecimal, allows, in a stream of received transport packets, to find the beginning of a transport package. There follows a byte 402 comprising an indicator to know that an error has been detected upstream and an indicator to know if the transport packet comprises a header of a packet of elementary stream or not. Then, an indicator gives a priority value for the packet, followed by 13 bits, distributed in bytes 402 and 403 which give an identification value of the transport packet, used for example in a transport stream comprising transport packets. carrying different types of streams (audio, video, ...). Byte 404 includes a two-bit flag to indicate the type of transport scrambling used, followed by two bits indicating the presence of an adaptation field according to the standard, and then four bits to hold the value of a counter. continuity, to know the sequence number for reassembly of an elementary stream packet if this elementary stream packet is too large to be put completely in a transport packet.

According to the example, the elementary stream packet header PES 41 comprises in the example illustrated 9 bytes. The first three bytes 411, 412 and 413 comprise a synchronization word or ("Packet Start Code Prefix" in English, for "packet start prefix code"), of a fixed value, making it possible to know a beginning of a packet of elementary streams in a stream. Follows a byte 414 which includes the "Streamjd" which is used to identify the type of elementary stream contained in the PES packet. According to the MPEG2-transport standard, a value of Streamjd identifies an elementary audio stream MPEG 1 or MPEG 2 The following two bytes 415 and 416 comprise a value representative of the length of the header PES, and make it possible to determine where the start of the payload of the PES packet. Next in bytes 417 and 418, various flags (or "flags" in English) indicating inter alia if the packet is scrambled ("scrambled" or "encrypted" in English), followed by byte 419 indicating the remaining length of the 'on your mind. In the example shown here, the remaining length is zero.

In this example, the headers 40 of the transport packet and 41 of the elementary stream packet are repeated only once for the three audio frames 30, 31 and 32. The frequency of occurrence of an elementary flow header at within a transport packet depends on the size of the payload of a transport packet and the size of the elementary stream packet. According to a variant of the invention, several elementary flow headers are present in a transport packet if the size of the payload of a transport packet allows it. According to another variant, an elementary stream packet is longer than a transport packet (for example 2048 bytes); a transport packet transports, according to the standard, the data of a single packet of elementary streams at a time. Figure 5 illustrates the encapsulation of several audio frames in a transport packet.

FIG. 5 diagrammatically illustrates the different levels of encapsulation of the data processed by the apparatus 2 embodying the invention, and more specifically two transport packets 50 and 51. The transport packet 50 (respectively 51) comprises:

a header 501 (respectively 511); and

a payload portion 502 (respectively 512) corresponding to a packet of elementary streams 5021 (respectively 5121). According to one variant, the payload is not limited to an elementary stream packet and comprises several elementary stream packets.

Alternatively, the payload comprises a portion of a packet of elementary streams.

The elementary stream packet 5021 (respectively 5121) comprises:

a header 50211 (respectively 51211); and

a payload portion 50212 (respectively 51212) comprising audio frames 502121, 502122 and 502123 (respectively 512121 and 512212. To illustrate the encapsulation method, the frame 512121 is the continuation of the frame 502123, which could not be put completely in the transport package 50.

Also, the frame 512212 is not complete and the data of the frame 512212 not present in the transport packet 51 are inserted into another transport packet which is not illustrated here.

As previously discussed, an elementary stream packet may be longer in length than a transport packet; in this case, the payload of an elementary stream packet is distributed in one or more transport packets, without repeating the header of the elementary stream packet.

FIG. 6 represents, in the form of an algorithm, a method for optimizing transmission of variable rate audio frames in a stream transport for transmission over a packet transport network implemented in the apparatus 2.

The method for optimizing transmission of variable rate audio frames in a transport stream for transmission over a packet transport network begins with a step 60 during which various variables necessary for its proper operation are initialized. Then, during a step 62, the device 2 receives a variable rate audio frame 262 (corresponding for example to one of the frames 30, 31 or 32). In a step 63, the apparatus 2 generates a time reference for each received audio frame (audio frame belonging to a set of received variable rate frames), allowing a time reference 263 to be associated with a reception timing. an audio frame.

In a step 641, the apparatus 2 encapsulates at least one variable rate audio frame in an elementary stream packet with the associated time reference 263, the elementary stream packet having a determined size 265.

During a step 642, at least a portion of the elementary stream packet is encapsulated in at least one transport packet (for example according to the format of the transport packet of FIG. 4.

In a step 65, at least a portion of a transport packet is transmitted, and step 62 is repeated.

According to the example described, the steps 641 and 642 are distinct. According to an alternative embodiment, the steps 641 and 642 are brought together in a single step 64; according to this variant, a received audio frame is directly encapsulated in a transport packet in accordance with the elementary stream format (without explicit creation of an elementary stream packet), and transmission of the transport packet is performed as soon as at least a determined part of the frame is complete. For example, as soon as the header of the transport packet is ready, the header of the transport packet is transmitted; then, as soon as the header of an elementary stream packet is ready, the header of an elementary stream packet is transmitted, and as soon as an audio frame begins to be read, the read data is transmitted. This transmission on the fly is possible with the method, all the information necessary for the construction of at least a portion of a transport packet being available upon receipt of an audio frame. This transmission on the fly is advantageous because the method does not require storing and / or receiving all the audio frame data that is inserted into a transport packet for transmission of the transport packet to begin. In fact, the method only requires the temporary storage of only one audio frame at a time, i.e., the last audio frame received. Thus, a delay between the moment of receiving an audio frame and the transmission of at least a portion of a transport packet that includes the audio frame is reduced. Advantageously, the transmission of each transport packet starts as soon as a transport packet header is constructed. Advantageously, the transmission of data corresponding to a packet of elementary streams, these data being encapsulated in a transport packet, begins as soon as a header corresponding to this elementary stream packet is constructed.

According to an advantageous embodiment of the invention, the determination of a size for the elementary flow packets comprises the respect of all or some of the following criteria, allowing an optimal filling rate of the packets:

the minimum size of the elementary stream packets excluding the header is equal to the maximum size of an audio frame; this so that the audio frames are too large to be contained in a single packet of elementary stream, because in this case it is necessary to distribute the audio frame in several packets of elementary stream, and thus several headers of elementary streams are necessary for the same frame this is not conducive to effective packaging because the number of bits needed to package ^• an audio frame and increases;

the minimum size of the elementary stream packets is an integer times the size of the out-of-box transport packets, a constraint that may be imposed by the encoding standard used, such as the MPEG2-system standard, which imposes that can not have multiple PES headers in the same transport packet. The determined size is equal to or greater than greater than the minimum values and equal to or less than the maximum value. If two criteria are included in the calculation of the minimum size of a packet of elementary streams, the maximum size is chosen between the two minimum sizes and rounded to a multiple of the payload size of a packet. transport. From the point of view of the efficiency of the packaging, it is desirable to include the least possible elementary flow stubbornness, but this has a cost on the zapping of a receiver receiving such a stream, because when this receiver changes flux during a zapping, it can begin the decoding only by an elementary stream header; if the transport stream comprises an elementary stream with very spaced elementary stream headers, it is necessary to wait relatively long before decoding of such a stream can begin.

According to an implementation variant of the invention, the maximum size of the elementary stream packets is determined by a maximum zapping time; according to this variant, the size of the elementary stream packets is limited to a value enabling a receiver to find, during a change of stream, an elementary stream packet header in a determined maximum time period (judged as acceptable by a user), for example within a maximum of 500 ms.

According to one variant, this determination of the maximum size of the elementary stream packets can be combined with the determination of the minimum size as described above. In such a case, the determined size of the elementary stream packets is equal to or greater than the largest of the minimum values and equal to or less than the maximum value.

According to one variant, the size of the elementary stream packets is an editable configurable parameter.

According to another variant, the size of the elementary stream packets is a pre-recorded and fixed configuration parameter.

According to another variant, the size of the elementary stream packets is a pre-registered and modifiable configuration parameter.

According to an implementation variant of the invention, a step of prior analysis of the characteristics of the flow of variable rate audio frames is performed; then, a value for the size of the elementary stream packets is determined according to the result of this analysis. Such an analysis makes it possible in particular to determine the characteristics such as the maximum bit rate of the frame stream and the maximum size of the audio frames supplied to the transmission optimization method. According to another variant embodiment of the invention, the determination of a value for the size of the elementary stream packets is performed on data on the stream of audio frames supplied to the method. transmission optimization, such as frame rate and maximum frame size. With this information, the maximum size of an audio frame is calculated, the minimum number of transport packets for encapsulating an audio frame derived from this size according to the following relationship:

MinNbTpForFrame = (MinSizePESHeader + (MaxAudioRebit / DurationAudioTram)) / SizeChargeUtilePackingTransport. in which: - MinNbTpPourTram expresses the minimum number of transport packets necessary for encapsulation of an audio frame;

- MinSizePESEnter expresses the minimum size of an elementary stream packet header; - SizeLoadUtilePackageTransport expresses the size of the payload of the transport packets, for example 184.

- MaxAudioRbit / DurationAudioTram gives the maximum size of the audio frame. MaxDetAudio is for example 128kbits / s, and TimeTramAudio is for example 10 milliseconds; MinNbTpPourFrame thus calculated is then rounded upwards, and is given the value of the integer times the value of the payload of a transport packet. For example, if MinNbTpForFrame gives a value of 134.67, and the payload size of a transport packet is 184, the final value of MinNbTpPourFrame is 184. The size of the elementary stream packet is then calculated as follows:

Size_PES = N * MinNbTpForWheel * SizeLoadUtilePackageTransport

The value of the parameter N makes it possible to reduce the bit rate of the transport stream even more, but this has an implication on the zapping time of a receiver of such a stream. N is minimal equal to 1 and can go up to

N = (ZappingTime / MaximumTimeAudioTrue).

For example, the ZapTime variable has a value of 0.5 to have an access point every 0.5 seconds, allowing a receiver to begin decoding after a maximum of 0.5 seconds. According to an alternative embodiment of the invention, the transmission of at least one complete transport packet is made upon receipt of a variable rate audio frame. This variant of implementation In particular, this work makes it possible to reduce the delay between the reception of an audio frame and the transmission of the transport packets.

According to another variant of implementation of the invention, the transmission of at least one complete transport packet is made upon receipt of several variable rate audio frames.

The number of audio frames received before the transmission of a transport packet can be calculated by observing the delay between the input and the output of an audio frame, taking into account the delay introduced by the addition of the flow headers. elementary and transportation. Figure 7 shows in detail an example of implementation of step 63.

Step 63 begins with a test 631, in which the apparatus 2 checks whether an elementary stream packet header (for example 50211 or

51211 of Figure 5) is necessary. An elementary stream header is needed each time a new elementary stream packet starts.

If during the test 631, an elementary flow header is useful or necessary, during a test 632, the device 2 checks whether the treated elementary stream packet starts with a start of an audio frame or not (by for example: the audio frame 502121 starts in the elementary stream packet 5021, and the audio frame 512121 does not start in the elementary stream packet 5121).

If so, the elementary stream packet starts with an audio frame start, and during a step 633, a time reference corresponding to the time of reception of the processed audio frame is retained and put in the header of the audio frame. elementary stream packet (eg 50211).

If not, the elementary stream packet does not start with an audio frame start and during a step 634, the time reference for the elementary stream header is determined as the value of the time reference associated with the the time of reception of the audio frame to which is added the duration of reception of the frame (for example, the reference of time to put in the header of the packet of elementary flow

5121 corresponds to the moment of reception of the frame 502123 plus the number of bits of this frame divided by the bit rate of the frame 502123). Step 63 ends at the end of steps 633 or 634 or if the result of test 631 is negative. Fig. 8 shows step 64 of encapsulating variable rate audio frames in transport packets without dissociating steps 641, 642, 643.

In a step 81, a transport packet is created if necessary, i.e., at the first execution of step 64, or if the transport packet being processed is full. At the end of this step, if a header has been created, it is ready and transmitted.

In a step 82, a PES packet header or elementary stream is added to the previously created transport packet, if necessary, i.e., at the first execution of step 64, or if the PES package being processed is full. The header comprises a time reference, which can be calculated according to the algorithm presented in FIG. 7. The header also comprises a value of the length of the elementary stream packet, having a determined value. At the end of this step, if a header is created, it is ready and transmitted.

In a step 83, the audio frame read is added wholly or in part to the current transport packet until the current elementary stream packet is completely filled, or until the audio frame read is completely processed, that is, added in its entirety to a packet of elementary streams.

During a test 84, the device 2 checks whether the current elementary stream packet is full or not. If the current elementary stream packet is full, step 82 is reiterated.

If the packet is not yet full, during a test 86, the device 2 checks whether the current transport packet is full.

If the current transport packet is full, in a step 85, a new transport packet is created. Step 83 is then repeated.

If the current transport stream packet is not yet full, during a test 87, the device 2 checks whether the audio frame whose processing is in progress is completely encapsulated. If this is the case, a new audio frame is read back to step 62. If the audio frame being processed is not yet fully encapsulated, a new transport packet is created in step 85 and step 83 is repeated.

According to a variant, the steps comprising transmissions provide the data ready to be transmitted to a transmission buffer or FIFO ("first in, first out"), which brings them together before send all or part of it. This allows to group parts of transport packets or entire transport packets to burst them (or "burst" in English) or, on the contrary, to spread their sending in time, according to the characteristics of the network on which the packets are transmitted. According to another variant embodiment of the invention, the steps comprising transmissions provide the data ready to be transmitted to a transmission buffer in order to be grouped and encapsulated in RTP transport packets according to the IETF RFC 3550 standard (for " Real-time Transport Protocol "in English, for" Real-time Transport Protocol ") allowing, at the reception, to correct the delays of transmission occurred (jitter in French or" jitter "in English) and / or to deliver packets in sequential order for packets that are not received in order.

The invention also allows the reception of transport packets constructed according to the formats described above. The reception is for example implemented in one of the receivers 108 or 114. The structure of these receivers is similar to that of the apparatus 2, the interfaces being adapted to the reception of transport packets and to the reference supply. time and audio frames to an application, and the programs being adapted to the implementation of the receiving method.

Figure 9 illustrates a particular embodiment of a method of receiving a transport stream constructed according to the transmission method. The method begins with a step 90 during which the various variables useful for the implementation of the method are initialized.

Then the method includes receiving 91 of at least a portion of a transport packet.

Then, the receiver retrieves a time reference from the transport packet (step 92). Then, it extracts 93 of at least one variable rate audio frame from the received transport packets. This operation is dual of the operation or encapsulation operations performed by the apparatus 2.

In general, the receiver is suitable for reconstructing audio frames transmitted by the source and encapsulated in transport packets. According to variants, it implements a different coding.

According to various embodiments, the extraction of the audio frames begins before the reception of the entirety of a transport packet including the audio frame (s) extracted, thereby reducing latency.

Of course, the invention is not limited to the embodiments described above. In particular, several of the steps of the method can be executed in parallel, such as frame reception, encapsulation, and transmission, by adding communication means and buffer zones between these steps. This has the particular advantage of allowing a separation of specific tasks. In particular, the method can be implemented not by a single device, but by a set of separate devices.

In particular, the invention is not limited to the audio frame formats, elementary stream packet format, and the transport stream packet format described; the method can be applied to other audio frame formats, other basic packet formats, and other transport packet formats.

Several of the advantages provided by the invention can be illustrated by the following application example. The invention can be particularly advantageous in an environment of transmissions subject to high maximum throughput constraints, such as television transmission to mobile phones. In such an environment, a transport stream comprising audio and video typically should not exceed a few hundred kbit / s (kilobits per second). Thus, the impact of the audio component on the total bandwidth of the transport stream is relatively high, the audio component typically occupying a bandwidth of the order of 15 to 20% of the total bandwidth (audio + video). Under the strong constraint of maximum authorized bit rate, the video streams entering a mobile phone transmission platform are therefore significantly reduced in bandwidth before transmission. In this context, the optimization of transmission of variable rate audio frames in a transport stream according to the transmission optimization method of the invention makes it possible to transmit the video with a better quality by making it possible to reduce the bandwidth beforehand. transmission by the gain obtained on the efficiency of the packaging of the audio stream, while making no concession on the quality of the audio stream.

Claims

A method of optimizing transmission of a variable rate audio frame stream in a transport stream for transmission over a packet transport network, characterized in that said method comprises the following steps: receiving at least one audio frame (62) of variable bit rate; generating a time reference (63) for each audio frame belonging to a set of received variable rate audio frames, said time reference being associated with a reception timing of said audio frame; encapsulating at least one variable rate audio frame in an audio elementary stream packet (641) with the generated time reference, said elementary audio stream packet having a size determined according to the characteristics of said stream audio frame stream variable and according to a payload of transport packets of said transport stream; encapsulating at least a portion of the audio elementary stream packet in at least one transport packet (642); and - Transmission (65) of at least a portion of a transport packet. 2. Method according to claim 1, characterized in that the size of said elementary flow packet is determined according to at least one of the following criteria: a first minimum size of said out-of-box elementary stream packet is equal to the maximum size of the audio frames of said audio frame stream, a second minimum size of said elementary stream packet is an integer times the size of a transport packet without the header of said transport stream, a third maximum size of said elementary stream packet as a function of a maximum decoding delay of said transport stream, and that when the at least one of the criteria comprises said first and second minimum sizes, said size of said elementary stream packet is equal to or greater than the largest of said first and second minimum sizes and when the at least one of the criteria includes said third maximum size, the size of said elementary stream packet is less than or equal to the third maximum size. 3. Method according to any one of claims 1 to 2, characterized in that the method comprises a step of prior analysis of the characteristics of said stream of variable rate audio frames and said size of said packet of elementary audio stream is determined in depending on the result of this analysis. 4. Method according to any one of claims 1 to 3, characterized in that said size of said audio elementary stream packet is determined from characteristics of said stream of audio frames supplied to the transmission optimization method. 5. Method according to any one of claims 1 to 4, characterized in that the step of encapsulation of at least one audio frame, and the encapsulation step of at least a part of the elementary stream packet audio are gathered in a single step of encapsulating variable rate audio frames in transport packets. 6. Method according to any one of claims 1 to 4, characterized in that the step of encapsulation of at least one audio frame, and the encapsulation step of at least a portion of the elementary stream packet audio are distinct. 7. Method according to any one of claims 1 to 6, characterized in that said step of generating a time reference comprises the following steps: when the payload portion of an audio elementary stream packet begins with the start of an audio frame, the time reference for a header of an audio elementary stream packet is equal to the time reference corresponding to the moment of reception of the audio frame; and - when the payload packet elementary audio stream part does not start with the start of an audio frame, the reference of time for said header of a current audio elementary stream packet is equal to the sum of a corresponding time reference at the time of reception of the start of the audio frame and the reception time of the audio frame.