View Profile
View Forum Posts
Private Message
Hi. I have a very annoying problem with a video file: when I extract (demux) the original audio using Avidemux or ffmpeg, the resulting AAC duration is 06:09 instead of 06:12 (the duration of the video). Same thing (06:09) if I'm importing the video in Audacity.
When I import that original video in DaVinci Resolve, the audio has gaps seemly random, so it stays in sync. Is this some non-standard AAC? How can I extract the original 06:12 in duration audio?
The original file is fairly big 774MB: https://mega.nz/file/W0J3kSwA#kXavtGTVM8pFvdLSwkSwxRRm9rA2TgvSr-N8GSufk8E
I need the original audio to remux it in a processed version of that video. Or at least a WAV with the correct duration and no gaps in it.
+ Reply to Thread
Results 1 to 30 of 32
-
Last edited by elektro; 19th Aug 2025 at 12:13.
-
I encoded a new file from Avidemux (using wav/pcm audio into MKV) and the resulting file is normal,
no sync problems start to finish.
The original audio in your source does have some drop outs (approx 35 milliseconds) that are audible
but they do not affect the length -
Thanks. Perhaps those drop outs (the gaps that I mentioned when I imported the file in DaVinci Resolve) cause the audio to have a shorter duration when exporting with other programs. Is there a way to automatically fill in those gaps? Maybe with the app from your screenshot (iZotope RX).Originally Posted by davexnet
-
Here are 2 audios I did with FFMPEG.
They are the correct legnth.
See if either or both will work for you.
Just change to .mp4 if you want that format.Code:ffmpeg -i Bring_Me_the_Horizon.mp4 -vn -sn -c:a aac_at -aac_at_quality 0 -aac_at_mode cbr -b:a 192k -ac 2 Bring_Me_the_Horizon.mka
-
Something odd going on here. I did a Quick Stream Fix on the MP4 in VideoRedo. It removes a lot of video frames and the resulting file is 6min ~9 seconds long, still in sync. I used VDub to extract the audio into WAV from the VideoRedo QSF file. It's attached.
I therefore think the Audacity audio is correct (6min 9sec) and the video is corrupted.Last edited by Alwyn; 19th Aug 2025 at 21:16.
-
That was my conclusion too.Originally Posted by Alwyn
-
Here's a recoded version (639MB).
https://drive.google.com/file/d/15s-j_knbggJSz4C8Y_cUk5e-aZDcoW_5/view?usp=sharing -
If you remux into mkv (e.g. mkvmerge, or mkvtoolnix), you can use eac3to to fix the aac gaps (eac3to does not support mp4) . The audio 06:12.117 in audacity after fixing gaps
Code:command line: "eac3to" "Bring Me the Horizon feat. Aurora - liMOusine (Summer Sonic Festival, Japan, 18-08-2024) original.mkv" 2: eac3to.aac ------------------------------------------------------------------------------ MKV, 1 video track, 1 audio track, 0:06:13, 60i /1.001 1: MPEG2, 1080i60 /1.001 (16:9) 2: AAC, 2.0 channels, 48kHz, -100ms [a02] Extracting audio track number 2... [a02] Applying AAC delay... [a02] A remaining delay of +6ms could not be fixed. [a02] Creating file "eac3to.aac"... [a02] Audio has a gap of 21ms at playtime 0:00:02. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:07. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:08. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:09. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:10. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:11. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:12. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:13. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:14. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:15. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:18. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:19. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:20. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:00:20. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:21. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:22. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:24. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:00:26. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:00:28. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:30. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:31. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:36. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:37. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:00:40. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:46. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:48. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:48. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:00:51. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:52. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:00:53. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:02. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:05. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:12. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:14. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:19. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:23. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:26. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:28. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:28. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:29. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:29. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:32. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:34. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:35. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:35. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:36. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:43. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:51. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:52. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:52. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:56. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:57. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:01:57. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:58. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:59. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:01:59. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:00. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:05. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:08. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:10. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:12. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:14. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:15. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:15. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:16. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:18. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:20. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:21. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:23. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:25. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:26. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:28. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:29. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:30. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:33. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:02:34. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:34. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:36. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:41. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:02:59. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:03:07. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:13. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:14. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:16. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:18. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:20. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:22. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:03:34. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:35. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:35. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:03:38. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:38. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:39. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:03:40. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:42. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:03:57. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:59. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:03:59. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:07. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:16. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:17. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:21. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:22. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:23. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:29. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:29. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:32. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:37. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:41. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:44. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:46. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:48. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:49. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:54. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:54. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:04:56. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:04:59. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:09. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:14. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:05:16. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:17. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:18. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:19. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:05:20. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:21. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:23. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:25. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:26. <WARNING> [a02] Audio has a gap of 22ms at playtime 0:05:29. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:30. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:31. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:32. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:33. <WARNING> [a02] Audio has a gap of 21ms at playtime 0:05:35. <WARNING> [a02] Starting 2nd pass... [a02] Extracting audio track number 2... [a02] Applying AAC delay... [a02] A remaining delay of +6ms could not be fixed. [a02] Realizing AAC gaps... [a02] Creating file "eac3to.aac"... Video track 1 contains 9582 frames. eac3to processing took 4 seconds. Done.
-
That's interesting . Does eac3to remove the gaps and push the pieces together thus shortening the length?
I noticed it'a almost a second shorter, 47875 samples -
Any analysis of the two I posted in post#4 :
It was funny that when all I did in FFMPEG was change the format to acc & the file was then 06:09.
Using the code I posted. -
I note the original file is MPEG 2. MPEG 2 and AAC in an MP4 container...
-
You can use clever FFmpeg-GUI to create a correct (in length) PCM audio file.
1) Load your mp4 from post #1, click main page, click extract streams, set the aac audiostream and click extract.
2) Go back to source and drag the extracted aac into the window.
Click main page, click encode audio stream, select pcm as encoder, set length only (at change length & pitch), select preset custom, set all like in the pic, click OK and click encode. Done.
[Attachment 88352 - Click to enlarge]
3) If you want check the result, you can mux it with the videostream.
To do so, click multiplex, drag your mp4 into the window, on the left deselect all streams, then select only the videostream and the new created wav (pcm) audiostream, set your output container and click multiplex. Done.Last edited by ProWo; 20th Aug 2025 at 12:03. Reason: sample added
-
Originally Posted by davexnet
In 3.x versions, you can choose the behaviour for gap repair by using -silence or -loop
https://www.videohelp.com/software/eac3to/version-history
Code:v3.00 . . . * for gaps, edits & repairs > 1000ms eac3to now inserts silence by default * for gaps, edits & repairs < 1000ms eac3to now loops audio by default * option "-silence" forces eac3to to insert silence instead of looping audio * option "-loop" forces eac3to to loop audio instead of inserting silence . . .
The MP4 has variable timestamps - entries such as 14.985015 fps - they probably are related to the audio gaps . The average framerate is 25.732
The content is actually 23.976 telecined . If you honor flags, IVTC, the duration is 06:12 and matches the fixed audio. That's the way I would do it so it plays back properly everywhere , progressively CFR. The OP mentioned remuxing with "processed version of that video", not sure how that was done
The benefit of keeping AAC vs WAV is smaller filesize. There might be other types of gap repairs that are more advanced now (e.g. "AI", or machine learning) , but eac3to has been used for many years for gap repairs (BD streams, broadcast streams) -
There is no gap in the audio stream.Originally Posted by poisondeathray
If you export it and view it in an audio editor, you can see it.
The whole problem is a failed mux of the two streams.
If you simply mux the mp4 file to mkv, you will see that the video is a normal 29.97 fps stream and the audio stream is simply shorter.
This mp4 monstrosity was created in an attempt to align the two. -
To be clear, the eac3to "gaps" refer to the audio relationship with the videoOriginally Posted by ProWo
Not really "failed" because the MP4 is in syncThe whole problem is a failed mux of the two streams.
Yes the audio stream is shorter - that's the reason for this thread .If you simply mux the mp4 file to mkv, you will see that the video is a normal 29.97 fps stream and the audio stream is simply shorter.
This mp4 monstrosity was created in an attempt to align the two
If you remux with ffmpeg or mkvmerge, look at timestamps - they are variable. The average frame rate is ~25.736 FPS ie. the video slows down in sections to keep sync with the shorter audio . There are sections where it goes down to ~15fps
There are differences in in stretching the audio linearly vs. gap repair at specific spots. The end duration will be the same (or very close), but the middle parts might be slightly more out of sync . You'd have to test which method is better in this case, and that would be dependent on the history of the stream, how it was created -
I knew something is fishy with this file. I have another video with the same issues from the same TV station.
Thanks for reply. I must process the video (deinterlace with QTGMC + normalize the audio), that's why I need both the video and audio. This fixed video you posted had the frame removal done in a lossless way?Originally Posted by Alwyn
Indeed. A dumb idea. But if you mess with this video, even if you move it into a MKV container, the A/V sync is lost.I note the original file is MPEG 2. MPEG 2 and AAC in an MP4 container...Last edited by elektro; 20th Aug 2025 at 11:56.
-
It looks like audio came from a 24 fps source and the video from a 23.976 fps source (24 fps film slowed to 23.976 fps and telecined). Then the timestamps were manipulated to make the audio running time match the video running time.Originally Posted by ProWo
-
Basically I want to use Avisynth to deinterlace the video with QTGMC + amplify the audio level with Audacity and save the results as HEVC + ACC in a MKV container. That's why I need to demux both video and audio in the correct way. Simply demuxing V/A and remuxing them in MKV results in a heavy out of sync audio.Originally Posted by poisondeathray
Originally Posted by davexnet
In 3.x versions, you can choose the behaviour for gap repair by using -silence or -loop
https://www.videohelp.com/software/eac3to/version-history
Code:v3.00 . . . * for gaps, edits & repairs > 1000ms eac3to now inserts silence by default * for gaps, edits & repairs < 1000ms eac3to now loops audio by default * option "-silence" forces eac3to to insert silence instead of looping audio * option "-loop" forces eac3to to loop audio instead of inserting silence . . .
The MP4 has variable timestamps - entries such as 14.985015 fps - they probably are related to the audio gaps . The average framerate is 25.732
The content is actually 23.976 telecined . If you honor flags, IVTC, the duration is 06:12 and matches the fixed audio. That's the way I would do it so it plays back properly everywhere , progressively CFR. The OP mentioned remuxing with "processed version of that video", not sure how that was done
The benefit of keeping AAC vs WAV is smaller filesize. There might be other types of gap repairs that are more advanced now (e.g. "AI", or machine learning) , but eac3to has been used for many years for gap repairs (BD streams, broadcast streams)
Reading this thread made me confused a bit. I don't get what is actually wrong and needs fixing: the video, the audio or both? -
Then what do you suggest to do in order to properly demux the A/V streams and edit the video in Avisynth? If I do IVTC to get 23.976fps and then AssumeFPS(24) the video duration is still only slightly lower than 06:12.Originally Posted by jagabo
It looks like audio came from a 24 fps source and the video from a 23.976 fps source (24 fps film slowed to 23.976 fps and telecined). Then the timestamps were manipulated to make the audio running time match the video running time.Originally Posted by ProWo
Last edited by elektro; 20th Aug 2025 at 11:43.
-
Oops, sorry, my calculation was wrong. I accidentally used 3600+12 minutes instead of 360+12 minutes. So you're right adjusting the video to 24 fps doesn't reduce the running time by ~3 seconds.
-
Video - do not deinterlace the video, because the content is not interlaced . It's 23.976p content telecined . You should inverse telecine to get back the original unique frames
Only the audio needs to be fixed, because it's shorter . eac3to is one way of doing it by adjusting gaps, stretching and resampling is another. Withing stretching, there are various algorithms, some are better than others. You'd have to test to see which is better for your source
This uses the eac3to audio in post #8
This is just to preview sync e.g. mpchc . Looks ok, sounds ok in terms of av sync, duration is correctCode:ffVideoSource("Bring Me the Horizon feat. Aurora - liMOusine (Summer Sonic Festival, Japan, 18-08-2024) original.mp4",rffmode=1) tfm(pp=0) tdecimate() vid=last aud=bsaudiosource("eac3to.aac") audiodub(vid,aud) #info
If you're amplfying the audio in audacity, you're re-encoding anyways, unless you use a lossless metadata method like replaygain/aacgain . Not all players support that method
https://www.videohelp.com/software/AACGain
eac3to can be used to amplify audio as well in that 1st step -
I finally picked a solution. I think Alwyn was right: the video is actually messed up and the audio is right. I used VideoRedo (like he recommended) to fix the video and remove the useless frames that made the video longer than the audio. Now both the audio and video are in sync. That's the Avisynth script I used to process the file and then convert it to HEVC + AAC.
And the result if you're interested to check it out: https://mega.nz/file/b5pSWCDZ#UaEhk63eMupAy4CQnUunZe3h_6yCFqet0k1cK5Pu0t4Code:SetFilterMTMode("QTGMC", 2) v=D2VSource("Bring Me the Horizon feat. Aurora - liMOusine (Summer Sonic Festival, Japan, 18-08-2024) original_track1_[und].d2v") a=LWLibavAudioSource("Bring Me the Horizon feat. Aurora - liMOusine (Summer Sonic Festival, Japan, 18-08-2024) original_track1_[und]_DELAY -31ms.wav") AudioDub(v, a).DelayAudio(-0.031) # in seconds AssumeTFF() AssumeFPS(29.970) TFM(slow=2, pp=7, clip2=QTGMC(preset="Slow", FPSDivisor=2, Sharpness=0.5)).TDecimate() # IVTC Deblock_QED(quant1=35, quant2=35) TTempSmooth() # light denoising Prefetch(4)
@poisondeathray: Thanks for trying to help but apparently the original audio didn't have gaps. Anyway, that's useful knowledge in case I have to deal with audio gaps (and I recently did and I fixed them by manually copy pasting small parts into the gaps using Audacity). -
I ran this on the original MP4:
and the video came out at 6min 8seconds (at 23.976fps). That closely matches the Audacity length. Audio sync spot on.Code:v=LSMASHVideoSource("H:\Videohelp\elektro\Bring Me the Horizon feat. VRD QSF Recode.mp4") a=LSMASHAudioSource("H:\Videohelp\elektro\Bring Me the Horizon feat. VRD QSF Recode.mp4") AudioDub(v,a) ConvertToYV12() TFM() Tdecimate() Prefetch(20)Last edited by Alwyn; 21st Aug 2025 at 08:45.
-
The problem with that method is there are dropped frames, jumps in the motion; some stutters and duplicates when you compare it to the other method.Originally Posted by elektro
There are single frame drops, but there are some obvious jumps where there are more than consecutive frame missing
1697-1698 (2 frames missing)
3246-3247 (2 frames missing)
6082-6083 (2 frames missing)
6083-6084 (2 frames missing)
7025-7026 (2 frames missing)
single drops
3111-3112
6389-6390
6837-6838
8842 missing
duplicates
650-651
4429-4430
5247-5248
6080-6081
6674-6675
7024-7025
7570-7571
7840-7841
This isn't the full list, but there should be 84 drops if you do the math, not including the duplicates
It's not "audio gaps" per se - the term is relative to the video timestamps . Since the audio is shorter , the timestamps dictate video speed changes (slows down in sections) to keep sync in the original video@poisondeathray: Thanks for trying to help but apparently the original audio didn't have gaps. Anyway, that's useful knowledge in case I have to deal with audio gaps (and I recently did and I fixed them by manually copy pasting small parts into the gaps using Audacity).
ie . Instead of fixing the audio, you're now shortening the video by dropping good frames to match the shorter audio . I don't know if eac3to method is the "best" way to handle the audio, but I personally wouldn't be dropping good frames and causing jumps and stutters -
I did a dupe check using DupMC on my script in post #23 and it didn't come up with any dupes.
-
Ok, I'll try that method too. I have another video with the exact same issue, so I have to pick the right method.Originally Posted by poisondeathray
The problem with that method is there are dropped frames, jumps in the motion; some stutters and duplicates when you compare it to the other method.Originally Posted by elektro
There are single frame drops, but there are some obvious jumps where there are more than consecutive frame missing
1697-1698 (2 frames missing)
3246-3247 (2 frames missing)
6082-6083 (2 frames missing)
6083-6084 (2 frames missing)
7025-7026 (2 frames missing)
single drops
3111-3112
6389-6390
6837-6838
8842 missing
duplicates
650-651
4429-4430
5247-5248
6080-6081
6674-6675
7024-7025
7570-7571
7840-7841
This isn't the full list, but there should be 84 drops if you do the math, not including the duplicates
It's not "audio gaps" per se - the term is relative to the video timestamps . Since the audio is shorter , the timestamps dictate video speed changes (slows down in sections) to keep sync in the original video@poisondeathray: Thanks for trying to help but apparently the original audio didn't have gaps. Anyway, that's useful knowledge in case I have to deal with audio gaps (and I recently did and I fixed them by manually copy pasting small parts into the gaps using Audacity).
ie . Instead of fixing the audio, you're now shortening the video by dropping good frames to match the shorter audio . I don't know if eac3to method is the "best" way to handle the audio, but I personally wouldn't be dropping good frames and causing jumps and stutters
I'm using Eac3to and More GUI. What options to pick? Or what is the exact command line to fix the audio? I don't want to make the audio worse.Last edited by elektro; 21st Aug 2025 at 09:59.
-
Video is longer than audioOriginally Posted by elektro
Your general options are to adjust the video to match audio, or audio to match video - and that applies to everything, not just this video
For video to match the audio as-is , you can (A) speed it up after proper IVTC (I suspect videoredo is dropping good frames, I wouldn't use it in this case) , or (B) resample the video using interpolation / optical flow / machine learning (C) is what you are doing now - dropping frames or duplicating frames to match the audio . (D) is what the original is doing - using timestamps to adjust "on the fly" during playback
The problem with (B) is interpolation is not perfect and prone to artifacts - that video has sections fence/mesh backgrounds - those types of repeating patterns do not interpolate well . The problem with (A) is you end up with some non standard rate like 24.175, but at least you keep the original frames and motion is good if you've done it properly. The problem with (D) is some players might not play it back correctly
For audio to match video length, you can 1) try the eac3to gap repair , or 2) stretch - prowo showed 1 method above of lengthing the audio, but there are supposedly higher quality methods in dedicated audio software using better algorithms than what ffmpeg/libavfilter/sox use. I think there are machine learning ones now too that are probably better. Not sure which is best for this case in terms of the audio, you'd have to test it out.
There is a chance that any of these methods might be slightly out of sync in parts in the middle (the start and end should align) , depending on the history of the original video, and etiology of the A/V length discrepancy in the first place (ie. why is audio shorter? ) . e.g if original problem was linear, then a linear fix would be ideal. They all have pros/cons, you have to decide which is best for your case
The commandline was posted in the log above - it automatically process it with loops instead of silence for <1000ms since the discrepancies detected were 21msI'm using Eac3to and More GUI. What options to pick? Or what is the exact command line to fix the audio? I don't want to make the audio worse.
The audio will be technically worse, because little loops are inserted to extend the length. It sounds ok to me, but I'm not an audio engineer. So you should try the stretch methods and compare. Stretching would be technically worse too, since you're resampling the audio. If you don't want to touch the audio, then adjust the video instead... but dropping and duplicating frames is the worst method IMOLast edited by poisondeathray; 21st Aug 2025 at 10:21.
-
^ Thanks for the lengthy and logical explanation. It seems none of the options is ideal but I'll have to pick the one with less glaring issues.
I tried the eac3to method but now I'm trying to make that program work properly. I moved the mp4 to mkv with MKVToolNix then I used this command:
But the result has a 6:09 duration and it didn't detect any gaps.Code:eac3to "Bring Me the Horizon feat. Aurora - liMOusine (Summer Sonic Festival, Japan, 18-08-2024) original.mkv" 2: eac3to.aac
The log:
Also if I want to use options like -normalize I get errors. eac3to -test gives the error Nero Audio Decoder (Nero 6 or older) doesn't seem to be installed. I put the files of Nero AAC Codec in eac3to folder and same result. It seem this eac3to doesn't have all required dependencies. Do you have an idea what's wrong?Code:Running in fast mode Keeping dialnorm analyze: 100% MKV, 1 video track, 1 audio track, 0:06:13, 60i/1.001 1: MPEG2, 1080i60/1.001 (16:9) 2: AAC, 2.0 channels, 48kHz, dialnorm: 0dB, -100ms a02 Extracting audio track number 2... a02 Applying AAC delay... a02 A remaining delay of +6ms could not be fixed. a02 Creating file "eac3to.aac"... process: 100% Video track 1 contains 9582 frames. eac3to processing took 2 seconds. Done.
Last edited by elektro; 21st Aug 2025 at 11:10.
-
On this computer the eac3to version used was 3.27 . There might be some different default settings - eg.. "Running in fast mode" might omit the check . A 2nd pass is required for the gap check, but there is no mention of 2nd pass in your log (it's default in that old version)
I can't recall the Nero configuration, it should be mentioned in the eac3to thread at doom9. I think a workaround would be to use -libav -
I'm using the latest version 3.52 of eac3to. I checked out the eac3to.ini and there was -fast as an option and I removed it. What I'm trying to do is the get the fixed audio (using loops) as wav and also normalized (as loud as possible without clipping). If I get the result as aac without the proper volume, then that's an extra encoding I'd like to avoid.Originally Posted by poisondeathray
Can you send me your eac3to folder as archived?Last edited by elektro; 21st Aug 2025 at 11:21.
Quote
Similar Threads
-
join aac + mp4 without reencode
By Clemens Ratte-Polle in forum EditingReplies: 3Last Post: 14th Oct 2023, 20:46 -
What audio container works well with .mp4/.mkv containers? FLAC or AAC?
By Guernsey in forum Newbie / General discussionsReplies: 1Last Post: 2nd Dec 2021, 00:18 -
I want to output to MP4 with AAC audio - what software?
By Drconfused in forum EditingReplies: 6Last Post: 11th Feb 2021, 19:56 -
AVCHD MTS to MP4 question on interlacing and AC3 or AAC
By TwoJ in forum Video ConversionReplies: 1Last Post: 23rd Jan 2021, 17:59 -
mp4 aac not playing / corrupted
By krecik in forum AudioReplies: 1Last Post: 15th Sep 2020, 17:24