Opened 13 years ago
Closed 10 years ago
#10848 closed Patch - Feature (Fixed)
[PATCH] OSD: Add a graphical display of audio level to cut list editor
Reported by: | Owned by: | Jim Stichnoth | |
---|---|---|---|
Priority: | minor | Milestone: | 0.28 |
Component: | MythTV - General | Version: | 0.25-fixes |
Severity: | medium | Keywords: | OSD audio graph |
Cc: | Karl Egly | Ticket locked: | no |
Description
When using the cut list editor to remove ads etc it's difficult to place the cut precisely when the program is faded in or out. In these cases it's common to cut the dialogue or retain voice-overs. It's also impossible to create a cut list for DVB radio recordings.
This patch adds a visual graph of the audio level +/-0.5 frame around the current paused frame.
NB this patch can also be applied to git master.
Attachments (1)
Change History (11)
Changed 13 years ago by
Attachment: | 0001-OSD-Add-a-graphical-display-of-audio-level-to-cut-li.patch added |
---|
comment:1 Changed 13 years ago by
Owner: | set to Jim Stichnoth |
---|---|
Status: | new → accepted |
comment:2 Changed 13 years ago by
Status: | accepted → infoneeded |
---|
comment:4 Changed 13 years ago by
Lawrence,
As far as I can tell, the visualizer data is pretty much random for AC3 audio (which is true for almost all of my recordings). Have you had any success with AC3?
comment:5 follow-up: 7 Changed 13 years ago by
No, not tried AC3. All my recordings are either mpeg2/mp3 with some mp4 if transcoded. The audio output device is to ALSA analog.
I had assumed that the audio data would be decoded 16-bit mono/stereo/multi-channel Am I to assume that this isn't the case. Does it matter if the output is digital? That's a bit of a pain if all the visualisers have to do their own decoding. Appears that MythMusic's visualisers assume 8/16, 1/2 channel PCM too.
Do you have a small sample that I could try? Coincidentally I wrote a script last week to import a clip into the recordings database.
comment:6 Changed 13 years ago by
Further investigation indicates that the audio data are correctly formatted with AC3 - I used an mp4 video with AC3 audio produced with ffmpeg. However, it appears that the timecodes for audio are unreliable such that when you seek back and forth you get different audio data for the same location. Maybe this is a consequence of a bad or missing posmap?
I couldn't get mythcommflag to rebuild the posmap for my mp4 video. It appears that avformatdecoder doesn't understand framing in H264 NAL streams, only annexe B.
comment:7 Changed 13 years ago by
Status: | infoneeded → assigned |
---|
Replying to Lawrence Rust <lvr@…>:
No, not tried AC3. All my recordings are either mpeg2/mp3 with some mp4 if transcoded. The audio output device is to ALSA analog.
That was a good clue. When setting the output to be stereo (via Audio Settings), 2-channel AC3 recordings produce plausible visualizations, at least when using the Spectrum visualizer. There are still some oddities to sort through -- it often appears that only the left half of the spectrum is active (could this be two-channel mono?), and some recordings show very low levels on the Spectrum visualizer relative to the volume on the TV.
That said, I also noticed that when you enable the Spectrum visualizer and enter the cutlist editor, the visualizer conveniently continues to be displayed. This means that the patch's functionality is already more or less present, though not in themeable form.
comment:8 Changed 13 years ago by
Maybe the audio channel format for multi-channel AC3 isn't being passed to the visualiser then. The patch accounts for the number of announced channels but processes only the 1st two. It would be straightforward to sum all channels.
It's nice to have confirmation of function from the spectrum visualiser but it's not as informative for real editing.
comment:9 Changed 11 years ago by
Cc: | Karl Egly added |
---|
comment:10 Changed 10 years ago by
Milestone: | unknown → 0.28 |
---|---|
Resolution: | → Fixed |
Status: | assigned → closed |
This appears to have been committed as [d40411ee9887057d08d408a43539ab4c848d7fe3] in July 2014. Closing as fixed in master.
Please post patches against master. We do not put new features into the -fixes branches by default policy.