On Friday 14 December 2018 00:16:24 Huang-Huang Bao wrote: > 90 samples are only 90/44100 s = 2ms , which is ignorable. Yes, this not a problem. Problem is that first 90 samples is just "garbage" or rather initialization data for predictor. And to prevent hearing such "garbage" when decoding stream, you need to drop them. Put aptX stream into aptX decoder, but drop first 90 samples of decoded output. Hearing noise or any other disrupting sound also for just 2ms is not something which you want to hear. This also applies when resetting decoder, e.g. after aptX parity check does not match. In most cases, first 90 decoded samples should be just silence (at least this is doing ffmpeg and android's proprietary encoders). But when you start decoding not from beginning but from existing stream, then it is a problem. Decoding from the middle of stream happens when some A2DP packets are lost and decoder needs to be reset and start decoding again. -- Pali Rohár pali.rohar@xxxxxxxxx
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ pulseaudio-discuss mailing list pulseaudio-discuss@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
