1、 Rep. ITU-R BT.2044 1 REPORT ITU-R BT.2044 Tolerable round-trip time delay for sound-programme and television broadcast programme inserts Context and rationale (Question ITU-R 35/6) (2004) CONTENTS Page 1 Scope 2 2 References and Bibliography 2 3 Terms and definitions . 3 4 Abbreviations and acronym
2、s 3 5 Causes of delay and echo . 4 5.1 Context 4 5.2 Historical background. 4 5.3 System design and configuration factors versus operating practices . 5 5.4 System design and configuration factors which cause delay and echo 6 5.4.1 Encode/decode processes. 6 5.4.2 Propagation delay. 6 5.5 Magnitude
3、of delay in various transmission media 7 5.6 Magnitude of delay in video codecs . 7 5.7 Magnitude of delay in audio codecs . 8 5.8 Circuit symmetry 10 5.9 Operating practices which cause echo 10 5.10 Sound to picture delay 10 6 Effects of delay and echo . 10 6.1 Audio delay and video delay 10 6.2 Va
4、riables affecting disturbance 11 6.3 Echoes and fusion . 11 6.4 Types of disturbance and their effects 12 6.5 Adaptation. 12 7 Summary. 12 2 Rep. ITU-R BT.2044 1 Scope This Report is intended to review the effects of delay and level of echo in an audio foldback loop in a broadcast production context
5、 It also reviews the effect of audio-video delay. It does not attempt to take into account the effect of additional reverberation and noise in the listening environment and assumes that there is no significant loss in signal quality from the monitoring system. 2 References and Bibliography 1 Techni
6、cal Report ETR 250 July 1996 Transmission and Multiplexing (TM); Speech commu-nication quality from mouth to ear for 3.1 kHz handset telephony across networks. European Telecommunications Standards Institute. http:/pda.etsi.org/pda/queryform.asp 2 Technical Report ETR 262 January 1996 Broadband inte
7、grated services digital network (B-ISDN); Asynchronous transfer mode (ATM); Video on demand (VOD) network aspects. European Telecommunications Standards Institute. 3 Technical Report ETR 275 April 1996 Transmission and Multiplexing (TM); Considerations on transmission delay and transmission delay va
8、lues for components on connections supporting speech communication over evolving digital networks. European Telecommunications Standards Institute. http:/pda.etsi.org/pda/queryform.asp 4 Recommendation ITU-R BT.1359 Relative timing of sound and vision for broadcasting. International Telecommunicatio
9、n Union (November 1998). 5 ETSI/GSM Recommendation 06.10 Version 3.2.0 February 1992 GSM Full Rate Speech Transcoding. European Telecommunications Standards Institute. http:/pda.etsi.org/pda/queryform.asp 6 DAVIS, P. 1995 A tutorial on MPEG audio compression. IEEE Multimedia, p. 60-74. 7 ONVURAL, R.
10、 1994 Asynchronous transfer mode networks: performance issues. Boston, Artech House. 8 Technical Report TR 100 815 V1.1.1 February 1999 Digital video broadcasting; guidelines for the handling of Asynchronous transfer mode signals in DVB systems. European Telecommuni-cations Standards Institute. 9 C.
11、S0014-0 Version 1.0 Enhanced Variable Rate Codec (EVRC) (CDMA2000 specification) Third Generation Partnership Project 2 (3GPP2) EIA/TIA (December 1999). 10 SIU-WAH WONG 1991 An Evaluation of 6.4 kbit/s Speech Codecs for Inmarsat-M System. IEEE, p. 629-632. 11 Recommendation ITU-R BT.1377 Labelling o
12、f video and audio apparatus throughput (processing) delay. International Telecommunication Union (1998). 12 EVEREST, F. A. 1994 The Master Handbook of Acoustics. TAB Books/McGraw-Hill. 13 CCITT Recommendation G.114 (1989) Mean one-way propagation time, Blue Book, Fasc. III.1. 14 CCITT Recommendation
13、 G.131 (1989) Stability and echo, Blue Book, Fasc. III.1. Rep. ITU-R BT.2044 3 15 ES 200 677 V1.2.1 Public Switched Telephone Network (PSTN) March 1998 Requirements for handset telephony. European Telecommunications Standards Institute. 16 LOCHNER, J. P. A. and BURGER, J. F. 1958 The subjective mask
14、ing of short time delayed echoes by their primary sounds and their contribution to the intelligibility of speech. Acustica, Vol. 8, 1, p. 1-10. 17 MEYER, E. and SCHODDER, G. R. 1952 ber den Einflu von Schallrckwrfen auf Richtungslokalisation und Lautstrke bei Sprache. Nachr. Akad. Wiss., Gttingen, 6
15、 18 CHURCH, S. On beer recommended maximum decode delay: 3 ms. Satellite phone systems there are a number of competing systems including ICO, Iridium, AceS, AMSC-TMI and Inmarsat. These systems can be classed as low-Earth orbit (LEO) (ICO, Globalstar, Iridium and Teledesic),) or geostationary Earth
16、 orbit (GEO) (Inmarsat, Satphone, ASC, Thuraya, APMT, EAST). A typical GEO system, Inmarsat Mini-M, uses advanced multi-band excitation (AMBE) encoding at 4.8 kbit/s, with encode/decode delays comparable with those of GSM and CDMA terrestrial mobile telephone systems. In the GEO case however, the di
17、stance delay tends to make the encode/decode delays insignificant. Studio MPEG codec. Audio for broadcast may be embedded with video using MPEG-2 encoding, or it may be sent on separate ISDN/ATM lines. If embedded with video, the audio will be synchronized to the video. If sent separately, the audio
18、 may be uncompressed, giving a negligible encode/decode delay, or it may be compressed. Compression algorithms vary between manufacturers, so minimum and typical delays are cited. Due to the nature of audio signals a minimum bit rate for a given audio quality would deviate by a factor of 10 or more
19、over time. In general time-slots with a large demand on local bit rate are very short and surrounded by time-slots with very low bit rate. Most state of the art audio coding schemes therefore do some averaging of the bit rate over time to provide a constant bit rate. Depending whether the buffer for
20、 this averaging is in the encoder or decoder side, the encoding or the decoding delay is larger. There are several options for MPEG-1 audio compression: Layer I, Layer II and Layer III. Layer III is also known as MP3. MPEG-2 in addition to Layer I, II and III, offers advanced audio coding (AAC). In
21、MPEG-4 AAC was chosen as the baseline codec for natural audio and extended by several new tools and functionalities. In the context of round trip delay the low delay (LD) version of AAC coding is the most prominent extension. Rep. ITU-R BT.2044 9 Minimum delays for MPEG-1 Layer I, II, III, from 6, 1
22、8 End-to-end delay (including ISDN channel) for a commercial available audio codec 18. End-to-end delay of current audio coding schemes. MPEG audio encoded with MPEG video will generally have the same delay as the video. Layer Target bit rate (kbit/s) Compression ratio Theoretical minimum delay (ms)
23、 AAC-LD 64 12:1 20 Layer I 192 4:1 19 Layer II 128 6:1 35 Layer II 64 12:1 59 Coding Bit rate (kbit/s) Sampling rate (kHz) Typical delay (ms) AAC-LD stereo 128 48 60 AAC-LD mono 64 48 50 AAC stereo 128 48 172 Layer III stereo 128 48 326 Layer II stereo 128 48 224 Layer II stereo 128 24 (half mode) 3
24、98 G.722 64 48 10 Codec Bit rate (kbit/s) Algorithmic delay with bit reservoir set to zero (ms) 100% workload, burst transmission(ms) 100% workload, continuous transmission(ms) 30% workload, burst transmission (ms) 30% workload, continuous transmission(ms) Layer 2 192 34 Not available Not available
25、Not available Not available Layer 3 128 54 118 142 107 131 MPEG-4 AAC 96 55 82 211 63 192 MPEG-4 HE AAC 56 129 184 361 145 322 MPEG-4 AAC-LD 128 20 33 44 24 35 10 Rep. ITU-R BT.2044 5.8 Circuit symmetry It should not be automatically assumed that the two half-loops in a programme contribution loop w
26、ill be identical in processing delay, path delay, overall delay or leakage. It is common practice in broadcasting to use a high-quality circuit on the outside broadcast programme line only, with a lower quality circuit on the foldback line. It is also common practice to use audio-only foldback, even
27、 on video contribution loops. Both of these factors will affect the overall loop delay and the perceptual disturbance from delay and echo. 5.9 Operating practices which cause echo While some delay is unavoidable in an audio circuit, echo is generally avoidable. Echo can be caused both by system desi
28、gn and by poor operating practices. Although good operating practices will not automatically ensure an echo-free loop, they will minimize the echo within a given system design. The effectiveness of good operating practice in controlling echo depends to a large extent on the inherent echo in a loop.
29、In a 2-wire circuit, the inherent echo is determined by system design. In a 4-wire circuit, echo caused by audio leakage from foldback lines to programme lines can be minimized by: using a “mix-minus” or “clean feed” to foldback, and/or using appropriate muting of foldback when microphones are open,
30、 either by system interlock if a manual talk button is used, or by using a voice-operated mute circuit (a “ducker”), and/or by using closed headphone monitoring or in-ear monitoring to minimize leakage into open microphones. 5.10 Sound to picture delay Delay between sound and picture has also been w
31、ell studied and standards 4 have been in place for some time. The main perceptual problem with this type of delay is loss of lip-sync on speech. The ITU standard currently specifies an acceptable delay between sound and picture of +25 ms (sound leads picture) to 100 ms (picture leads sound) as measu
32、red at the final programme source selection element (usually the video master control room). Resynchronizing separate sound and picture with varying delay is conceptually a simple task, although the equipment can be expensive. If the sound leads the picture, a simple inexpensive audio delay unit may
33、 be employed to correct the sync. If picture leads sound, the picture must be delayed: this equipment is generally more expensive, although such a facility may be available in an existing video server. 6 Effects of delay and echo 6.1 Audio delay and video delay Video communications loops have existe
34、d for some years in videoconferencing and in broadcasting but are a relatively new technology compared with audio communications loops. For this reason, the subjective effect of long delay in a video communications loop ( 1 frame) is not as well-studied as the subjective effect of long delay in an a
35、udio loop. Because of this there is a relative scarcity of information on the subjective effects of delay in video loops compared with that for Rep. ITU-R BT.2044 11 audio loops. This contribution is therefore based mainly on information available on the subjective effect of delay and echo in an aud
36、io loop. Until more detailed information becomes available on subjective effects of delay in video loops, audio delay and echo is assumed to be the primary cause of disturbance, regardless of whether it is the limiting factor in overall audio-video loop delay. 6.2 Variables affecting disturbance Aud
37、io delays have a number of disruptive effects on speech communications systems. The effect of the delay depends primarily on two factors: the length of the delay and the return loss in the loop 1, 3. 6.3 Echoes and fusion For delays below the threshold in Fig. 1, the human hearing system cannot dist
38、inguish between the two arrivals on most types of speech and music. The two sounds therefore appear to “fuse” and become one louder sound. This has a positive effect on both perceived loudness and on speech intelligibility. The fusion effect becomes progressively weaker as delay increases, particula
39、rly for delays over 30 ms. Where fusion does not occur, the echo becomes disturbing. Rap 2044-010 2040608010120242016128404812abDelay time (ms)Echo levelrelativetoprimary soundlevel(dB)Curves a Meyer and Schodder (primary sound level 55 Phons)b Lochner and Burger (primary sound level 50 Phons)Compar
40、ison of threshold of perception curves.FIGURE 1Audibility of delayed signal in the presence of direct signal, as a functionof delay and relative level (from 16)12 Rep. ITU-R BT.2044 6.4 Types of disturbance and their effects There are principally three types of disruptive effect from delay and echo:
41、 Long delay without significant return echo. This type of delay can be classified as a half-loop delay. This can cause difficulty in normal conversation, causing parties to talk over each other inadvertently and to break off sentences once they realize this has happened. This causes the conversation
42、 to be prolonged and broken and it is also distracting. Delays in satellite circuits ( 240-280 ms for each satellite link) are common causes of this type of disturbance. Echoes returning to the speaker. This type of delay can be classified as an integral multiple (1, 2, n) of a full loop delay. Thes
43、e interfere with the normal speech feedback mechanism from brain to mouth to ear to brain, causing stuttering and hesitation. Both the length of the delay and the level of the echo have a significant effect on the disturbance, the disturbance increasing with delay at a constant level and decreasing
44、with level at a constant delay. For long delays ( 240 ms) return losses as low as 50 dB can still cause some disturbance. Echoes returning to the listener. This can be classified as a 1 1/2, 2 1/2, etc. loop delay. These disturb the listener by reducing speech intelligibility. As with full loop dela
45、y, the disturbance increases with delay length. It is a common problem in public address systems where it is known as “slap echo” for single reflections and “flutter echo” for multiple reflections. The audibility of this type of delay is shown in Fig. 1, from 16. Echoes should be below this curve to
46、 be inaudible as separate sounds. The effects of echo delay and echo level on speech intelligibility and speech difficulty have been extensively studied by telecommunications authorities 1, 3, 14 and are relatively well understood. The effects on speech intelligibility are probably better understood
47、 however than the effects on speech difficulty. 6.5 Adaptation A learned or acquired tolerance to the effects of a stimulus or irritant is known as adaptation. Although some adaptation to echo-affected signals can be acquired, the amount of adaptation which is possible is generally quite limited. It
48、 can also take some time to adapt. The amount of adaptation which can be learned and the time taken to learn it is likely to vary considerably between individuals. Since much broadcast content, especially on live shows, is provided by guests rather than presenters, guidelines should be formulated on
49、 the assumption that no adaptation will occur in the available time. 7 Summary Looped communications systems are commonly encountered in broadcast production where multiple locations or remote locations are used, particularly in live programme production. The loop may be used to connect a studio interviewer and a remote guest or it may be used for foldback from the studio to a presenter at a remote location. Ideally, each separate stream of audio and
