Digigram IQOYA TALK, Руководство пользователя

Страница: 63 / 68

IQOYA TALK
Portable IP audio codec
63
APPENDIX B
Typical Latency Values
The latency you encounter for a connection between an IQOYA TALK and another IP codec depends on the audio
format chosen, the network and the functionalities that the devices provide - and use.
See the table below for maximum latency values in a half-duplex configuration, without failover mechanism or
FEC, with the buffer size set to 0, on an optimized network.
Audio type Audio formats Latency Description
PCM 24 bits 26 ms
MPEG 256 kbps 76 ms Layer II
AAC-LC 128 kbps 100 ms
AAL-LC + SBR 96 kbps 220 ms Aka HEv1
AAC-LC +SBR + PS 48 kbps 246 ms Aka HEv2
AAC-LD 160 kbps 56 ms
AAC-ELD 160 kbps 48 ms
All measurements were made on stereo samples at 48 kHz.
Network impact on latency
A basic principle: the more your network/architecture components are of quality, the less latency you will
encounter – nevertheless.
In case your network presents disturbances, IQOYA TALK allows you to:
- Absorb the network jitter by choosing an appropriate buffer size
- Choose a FEC scheme to make your stream resilient to packet loss with the AAC option, choose an AAC with
error resilience (making the coding itself more robust against errors).
However, keep in mind that each of these features will increase the overall latency of your link: The more features
you need to apply here and there to compensate for the suboptimal operation of a link in the chain, the more
processing and / or tape resources you need to use. This makes it not really efficient in terms of latency…
Impact of additional features on latency.
Basic principle: the more functionality you add, the less room you have for optimizing latency.
Having said all of this, here is the best scenario for low latency:
-
Half-duplex link without any fail-over!
-
In full-duplex, the latency will be slightly increased.