Bug #858
Missing audio in recordings and streaming via HTSP.
0%
Description
When recording from the danish terrestrial MUX1 (specifically "Mux1 kbhv-glx") no audio is recorded in the mkv files, nor is audio available when streaming via HTSP. However streaming from http://localhost:9981/stream/channelid/XXX works flawlessly, providing the audio stream. Upon starting a recording the following is output:
[INFO]:subscription: "DVR: Lyden af universet" subscribing on "DR2", weight: 300, adapter: "Conexant CX22702 DVB-T", network: "Mux1 kbhv-glx", mux: "Mux1 kbhv-glx: 474,000 kHz", provider: "DR", service: "DR2", quality: 100 [INFO]:dvr: DR2-Lyden af universet.2012-01-29.13-51 from adapter: "Conexant CX22702 DVB-T", network: "Mux1 kbhv-glx", mux: "Mux1 kbhv-glx: 474,000 kHz", provider: "DR", service: "DR2" [INFO]:dvr: # type lang resolution samplerate channels [INFO]:dvr: 1 H264 704 x 576 [INFO]:dvr: 2 AAC dan 96000 0 <disabled, no valid input> [INFO]:dvr: 3 TELETEXT [INFO]:dvr: 4 DVBSUB dan [INFO]:dvr: 5 DVBSUB dan
The related MUX2 ("Mux2 kbhv-glx-jydr") works without any issues.
History
Updated by Mikael Jepsen over 12 years ago
Same problem here. As a temporary workaround, I record to a ts file instead, which has audio. Would be nice to be able to use mkv though...
Søren Bøg wrote:
When recording from the danish terrestrial MUX1 (specifically "Mux1 kbhv-glx") no audio is recorded in the mkv files, nor is audio available when streaming via HTSP. However streaming from http://localhost:9981/stream/channelid/XXX works flawlessly, providing the audio stream. Upon starting a recording the following is output:
[...]
The related MUX2 ("Mux2 kbhv-glx-jydr") works without any issues.
Updated by Mikael Jepsen over 12 years ago
Found the reason for the missing audio:
The danish MUX1 channels send HE-AAC with explicit SBR signalling (AudioObjectType 5), which the latm parser does not recognize. In comparison, the danish MUX2 channels also send HE-AAC, but not with explicit SBR (AudioObjectType 2), which is why tvheadend parses the latter correctly.
When recording to a ts file, I believe the audio is not parsed, but simply forwarded without even looking at it, so that works fine as well.
Fix made and pull request sent, see https://github.com/andoma/tvheadend/pull/56
Updated by Mikael Jepsen over 12 years ago
Andreas has merged the changes (https://github.com/andoma/tvheadend/commit/b8778005b4d9c1cc173f970a67bdfb4f1d7550e8) - could you take a stab at testing them out and close the issue if they work?
Thanks,
Mikael