Hi everyone, I have a few questions about the time delay of audio coding, any answer is greatly appreciated, thanks very much.

Firstly, we will not take the communication into account, that means only audio coding precess is considered.

1. Is the time delay always happend when coded? I found there is time delay in some public listening test samples, contrarily, no delay is found when I coded the audios by myself.

2. If time delay occurs, what is the cause of time delay? Is it generated by codec, or the delay is added artificially?

3. If no communication error is taken into account, I mean the time delay is caused only by audio coding, then, the delay is constant or time-varying?

4. What methods can be utilized to give a better estimation of the time delay?

I hope anyone can help me to answer these questions, even one of them, thank you very much.

There are 2 possible delays: encoder delay, and algorithmic delay.

*encoder delay: an encoder not designed with real-time bi-directionnal communication in mind might want to want for a few more input frames before coding the current one. It might, as an example, do this in order to decide block type (attack detection), or to change the bits distribution between frames (look-ahead).

*algorithmic delay: overlapped transforms need some samples from the next frame before being able to compute the time->freq transform of the current frame.