[Libav-user] Using libavformat/libavcodec with sample accurate seeking

[Anton Shekhovtsov]

2016-08-28 8:16 GMT+03:00 Christopher Snowhill <kode54 at gmail.com>:

> I've noticed that when using these libraries in my own projects, at
> least in VGMStream, and with the WMA Pro codec, it is necessary to
> perform my own pre-roll when seeking for looping, or else the codec
> ramps the audio in from silence on every loop as it catches back up to a
> valid state. I currently use either two seconds, or the beginning of the
> file if it's less than two seconds prior to the intended seek point,
> then discard sample data that is returned by the decoder.
>
> Is this always necessary by design? Or am I misusing the seeking API?
>
> I currently use:
>
> int64_t ts = target_sample * formatCtx->duration / total_samples;
> avformat_seek_file(formatCtx, -1, ts - 1000, ts, ts, AVSEEK_FLAG_ANY);
>
>
> _______________________________________________
> Libav-user mailing list
> Libav-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/libav-user
>
>
I noticed similar behavior with AAC decoder, work around by dropping 1024
decoded samples. (1024 is arbitrary number that worked for me)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ffmpeg.org/pipermail/libav-user/attachments/20160828/55f56802/attachment.html>