separate audio: sync issues at DVinfo.net

Jim Newberry · June 8th, 2009, 10:19 PM

I've begun recording audio with a Microtrack II (at 48K). I use Mpegstreamclip to convert my 5D .mov files to 29.97 and audio to 48K. At that point, do I still need to slow the 5D and/or Microtrack audio to 99.9%?

Are some people getting audio sync issues even without separate audio?

John Benton · June 14th, 2009, 03:50 PM

I have had issues with mpegStramclip 29.97
(syncing sound recorder on the zoom H4)
99.9% speed in FinalCut does Not always work, so

Last time I converted to ProRes and left the time field Blank
now no matter what% I sync the H4 audio to it doesn't work
(99.9 / 99.98 / 99.97 even 100.01 / 100.02)

Nothing seem to work in either situation for me. Each clip seems to require an entirely unique setting.

Evan Donn · June 17th, 2009, 02:23 PM

I've been testing a bunch of different workflows today - I'm mostly syncing multiple clips from the 5D with continuous long takes from the H4n @ 16bit/48khz. The most consistent one I've found is actually not real intuitive:

1. convert to ProRes in MPEG Streamclip, don't change frame rate
2. Add all video clips to a new 30 fps FCP sequence, add all H4n audio to A3-4
3. Run pluraleyes to sync everything automatically
4. in each new synced sequence change audio speed to 99.9%
5. trim H4n audio with blade tool to match video clip lengths
6. delete original audio, link H4n audio with video clips
7. drag newly synced clips into a new bin
8. edit with the synced clips in a ProRes 29.97 sequence

Now this seems like a strange workflow because the final synced sequences from Pluraleyes contain a single long audio track and multiple corresponding video clips - so you would think each clip would at least start out in sync and then drift over the length of the clip, so retiming the audio would break sync for everything but the first clip. Doesn't work this way though... in the initial synced sequences each subsequent clip is more out of sync than the last, and changing the audio speed brings all of them back in sync. The longest sequence I've tested this on so far is about 25 minutes long.

I've tried conforming to 29.97 in cinema tools, syncing in 29.97 timelines, retiming the audio before using pluraleyes, etc - so far this is the only workflow that seems to work consistently for me.

Bill Binder · June 17th, 2009, 04:49 PM

What I find interesting here is that there's a lot of confusion about some of this.

First there is the problem with 5D2 audio being at 30.00 fps and not 29.97. That can cause problems when mixing with other video sources no doubt. But the thing it seems like a lot of people just aren't getting at all -- and I say this as someone who has been doing double system for music shows for years -- is that no two recorders will necessarily remain in sync without being jammed together or running off the same wordclock.

This can be true potentially even for the EXACT SAME make and model, and even on the most "pro" of models. Take two super high-end 744s, and don't jam sync or link up their word clocks, and even they in all likelihood will start drifting from one another eventually (given enough time) -- THAT'S why jamming and linking word clocks exist in the first place!

So, knowing that, someone like me essentially EXPECTS drift issues when syncing any external sound source to any other video source. It's a given, not a surprise, it's expected! Now, it's not to say you might get lucky, and have two units that can stay synced for long periods of time. Sure that's possible, but it's WAY MORE likely to go the other way around, to drift. With musical performances I almost always have to sync the head of the clip, then go out to the end (often an hour or more later) and stretch/shrink until the tail gets sync too. At that point, I know the entire clip is synced. Been doing it for years and years.

I've been saying this since the launch of the 5D2 but seems like people still expect perfect sync (and I'm talking regardless of the 44.1 vs. 48 issue and the 29.97 vs. 30.00 fps issue).

Anyway, just thought I'd mention it one more time...

John Benton · June 17th, 2009, 11:26 PM

Thanks Evan,
Thanks Bill,

Hmm, I will do this with important things. Otherwise, I'm gong to have to figure out an on a good on Cam mike

Which means I might use the H4 as an on Cam mic an plug it's headphone jack into the 5D
(to use either the XLR Or the stereo mics) on board...

I have an on board mic, but its not great & I get A LOT os HISS

anyone see problems with this?

Thanks
J

Evan Donn · June 18th, 2009, 12:14 PM

Quote:

Originally Posted by John Benton

Which means I might use the H4 as an on Cam mic an plug it's headphone jack into the 5D
(to use either the XLR Or the stereo mics) on board...

I have an on board mic, but its not great & I get A LOT os HISS

anyone see problems with this?

I've shot some that way, and the problem is the hiss has nothing to do with the mic or other audio source - it's a result of the camera's auto gain bumping up the signal. Running through the H4 I certainly get better audio than the in camera mic, but every time someone stops speaking the camera cranks up the signal and the hiss - and then as soon as they start speaking again you hear it clamp down on their first word. It sounds like the magic lantern firmware is going to be able to prevent this, so once that's available it's probably a good workflow.

Matt Gottshalk · June 18th, 2009, 07:55 PM

From the 5dcinema forum:

Superb 30p to 24p Workﬂow

1. Cut video in 30 fps timeline

2. Create copy of ﬁnished edit so you'll have two clips, one to convert to 24p and one to
preserve audio.

2. Open Cinema Tool (part of Final Cut Studio), File, "Open Clip", choose "Conform",
specify "23.98" to conform to.

Explanation: This keeps the same number of frames in the clip, it only changes the
timebase from 30fps to 24fps, essentially creates overcranked footage by 6 frames,
(desirable on some shots). As a result this makes the clip longer and slows the audio
down which is undesirable where audio is important. At this point you've already created
a copy of the original clip therefore preserving the original audio which we'll relink later
in Final Cut.

3. Now open the inspector window so we can do some tweaking. Under "Frame
Controls" tab, click the button to the right of "Frame Controls:" to enable this feature.
Once enabled, select "On" for "Frame Controls:" Underneath the "Retiming Control"
section is where we'll make changes. For "Set Duration to:" click the radio button where
it will allow us to put in a duration. NOW THIS IS THE MOST IMPORTANT PART, OPEN
THE COPY OF THE CLIP WE CREATED THAT STILL HAS THE 30 FPS TIMEBASE
AND PUT IT INTO THE FINAL CUT PRO TIMELINE, DELETE THE VIDEO TRACK
AND ENSURE THE TIMEBASE OF THE SEQUENCE IS SET TO 23.98, GO TO THE
END OF THE AUDIO AND READ THE TIMECODE, THIS WAY WE'RE ENSURING
THAT WE'RE RETIMING (SPEEDING BACK UP) THE CLIP SO IT WILL MATCH THE
ORIGINAL AUDIO. This number is the timecode you'll put into the duration ﬁeld,
(example01:08:01:06) then depending on how well you want the footage to look also
taking into consideration the time it will require to compress it choose the desired quality
under "Rate Conversion:". I choose, "Best (High quality motion compesnation)" to get
superb results, however this increases the compression time.
Important: Make sure that no matter which codec you use that it remains as a 23.98 fps
timebase.

4. Now take the treated clip into Final Cut, we've already set up the sequence from step
3. Add the clip to video track 1, relink and presto!

I've extensively tested this and ﬁnd it to be the best way to convert 30p to 24p for the
time being, or at least until and hopefully Canon comes up with a ﬁrmware to address
the issue.

Additional Tips: If you have multiple clips to treat, in Cinema Tools, choose "Batch
Conform" locate one ﬁle where all the ﬁles are located. Then setup batch conversions in
5dMkII 30P to 24p conversion.

Jim Newberry · June 18th, 2009, 08:05 PM

Quote:

Originally Posted by Evan Donn

I've been testing a bunch of different workflows today - I'm mostly syncing multiple clips from the 5D with continuous long takes from the H4n @ 16bit/48khz. The most consistent one I've found is actually not real intuitive:

1. convert to ProRes in MPEG Streamclip, don't change frame rate
2. Add all video clips to a new 30 fps FCP sequence, add all H4n audio to A3-4
3. Run pluraleyes to sync everything automatically
4. in each new synced sequence change audio speed to 99.9%
5. trim H4n audio with blade tool to match video clip lengths
6. delete original audio, link H4n audio with video clips
7. drag newly synced clips into a new bin
8. edit with the synced clips in a ProRes 29.97 sequence

Now this seems like a strange workflow because the final synced sequences from Pluraleyes contain a single long audio track and multiple corresponding video clips - so you would think each clip would at least start out in sync and then drift over the length of the clip, so retiming the audio would break sync for everything but the first clip. Doesn't work this way though... in the initial synced sequences each subsequent clip is more out of sync than the last, and changing the audio speed brings all of them back in sync. The longest sequence I've tested this on so far is about 25 minutes long.

I've tried conforming to 29.97 in cinema tools, syncing in 29.97 timelines, retiming the audio before using pluraleyes, etc - so far this is the only workflow that seems to work consistently for me.

Evan: I tried Plural Eyes and found it very helpful. Regarding step 4 above: do you convert all audio to 99%--both 5D audio and separate audio recorder (say if you wanted to mix both)?

Evan Donn · June 25th, 2009, 10:00 AM

Quote:

Originally Posted by Jim Newberry

Evan: I tried Plural Eyes and found it very helpful. Regarding step 4 above: do you convert all audio to 99%--both 5D audio and separate audio recorder (say if you wanted to mix both)?

No, just the audio from the H4n - the camera audio should always be in sync with the video, and you should hear that it's out of sync with the H4n audio. Once you retime the recorder audio they should fall back in sync, although I don't know that you'll really want to mix the two - there's always slight differences so you'll often hear some flanging if you play both at once.

Christopher Lovenguth · June 25th, 2009, 03:26 PM

I'm starting to think this is a FCP issue since I have no problem with sync from H4N (recording 48Hz 24bit) editing in Premiere Pro exporting out using uncompressed audio and ProRes.

However I hope someone can tell me why after this being successful, I have audio sync issues just converting this footage to 720p H.264 uncompressed audio every time?

June 8th, 2009, 10:19 PM	#1
Jim Newberry Regular Crew Join Date: May 2007 Location: Los Angeles, California Posts: 170	separate audio: sync issues I've begun recording audio with a Microtrack II (at 48K). I use Mpegstreamclip to convert my 5D .mov files to 29.97 and audio to 48K. At that point, do I still need to slow the 5D and/or Microtrack audio to 99.9%? Are some people getting audio sync issues even without separate audio?

June 14th, 2009, 03:50 PM	#2
John Benton Major Player Join Date: Oct 2005 Location: NYC Posts: 658	Ugh I have had issues with mpegStramclip 29.97 (syncing sound recorder on the zoom H4) 99.9% speed in FinalCut does Not always work, so Last time I converted to ProRes and left the time field Blank now no matter what% I sync the H4 audio to it doesn't work (99.9 / 99.98 / 99.97 even 100.01 / 100.02) Nothing seem to work in either situation for me. Each clip seems to require an entirely unique setting.

June 17th, 2009, 02:23 PM	#3
Evan Donn Major Player Join Date: Feb 2006 Location: Philadelphia Posts: 795	I've been testing a bunch of different workflows today - I'm mostly syncing multiple clips from the 5D with continuous long takes from the H4n @ 16bit/48khz. The most consistent one I've found is actually not real intuitive: 1. convert to ProRes in MPEG Streamclip, don't change frame rate 2. Add all video clips to a new 30 fps FCP sequence, add all H4n audio to A3-4 3. Run pluraleyes to sync everything automatically 4. in each new synced sequence change audio speed to 99.9% 5. trim H4n audio with blade tool to match video clip lengths 6. delete original audio, link H4n audio with video clips 7. drag newly synced clips into a new bin 8. edit with the synced clips in a ProRes 29.97 sequence Now this seems like a strange workflow because the final synced sequences from Pluraleyes contain a single long audio track and multiple corresponding video clips - so you would think each clip would at least start out in sync and then drift over the length of the clip, so retiming the audio would break sync for everything but the first clip. Doesn't work this way though... in the initial synced sequences each subsequent clip is more out of sync than the last, and changing the audio speed brings all of them back in sync. The longest sequence I've tested this on so far is about 25 minutes long. I've tried conforming to 29.97 in cinema tools, syncing in 29.97 timelines, retiming the audio before using pluraleyes, etc - so far this is the only workflow that seems to work consistently for me. __________________ My latest short documentary: "Four Pauls: Bring the Hat Back!" Last edited by Evan Donn; June 17th, 2009 at 02:56 PM.

June 17th, 2009, 04:49 PM	#4
Bill Binder Major Player Join Date: Jun 2005 Location: San Francisco Posts: 542	What I find interesting here is that there's a lot of confusion about some of this. First there is the problem with 5D2 audio being at 30.00 fps and not 29.97. That can cause problems when mixing with other video sources no doubt. But the thing it seems like a lot of people just aren't getting at all -- and I say this as someone who has been doing double system for music shows for years -- is that no two recorders will necessarily remain in sync without being jammed together or running off the same wordclock. This can be true potentially even for the EXACT SAME make and model, and even on the most "pro" of models. Take two super high-end 744s, and don't jam sync or link up their word clocks, and even they in all likelihood will start drifting from one another eventually (given enough time) -- THAT'S why jamming and linking word clocks exist in the first place! So, knowing that, someone like me essentially EXPECTS drift issues when syncing any external sound source to any other video source. It's a given, not a surprise, it's expected! Now, it's not to say you might get lucky, and have two units that can stay synced for long periods of time. Sure that's possible, but it's WAY MORE likely to go the other way around, to drift. With musical performances I almost always have to sync the head of the clip, then go out to the end (often an hour or more later) and stretch/shrink until the tail gets sync too. At that point, I know the entire clip is synced. Been doing it for years and years. I've been saying this since the launch of the 5D2 but seems like people still expect perfect sync (and I'm talking regardless of the 44.1 vs. 48 issue and the 29.97 vs. 30.00 fps issue). Anyway, just thought I'd mention it one more time...

June 17th, 2009, 11:26 PM	#5
John Benton Major Player Join Date: Oct 2005 Location: NYC Posts: 658	Thanks Evan, Thanks Bill, Hmm, I will do this with important things. Otherwise, I'm gong to have to figure out an on a good on Cam mike Which means I might use the H4 as an on Cam mic an plug it's headphone jack into the 5D (to use either the XLR Or the stereo mics) on board... I have an on board mic, but its not great & I get A LOT os HISS anyone see problems with this? Thanks J

June 18th, 2009, 07:55 PM	#7
Matt Gottshalk Major Player Join Date: Dec 2003 Location: Potomac Falls, VA Posts: 215	From the 5dcinema forum: Superb 30p to 24p Workﬂow 1. Cut video in 30 fps timeline 2. Create copy of ﬁnished edit so you'll have two clips, one to convert to 24p and one to preserve audio. 2. Open Cinema Tool (part of Final Cut Studio), File, "Open Clip", choose "Conform", specify "23.98" to conform to. Explanation: This keeps the same number of frames in the clip, it only changes the timebase from 30fps to 24fps, essentially creates overcranked footage by 6 frames, (desirable on some shots). As a result this makes the clip longer and slows the audio down which is undesirable where audio is important. At this point you've already created a copy of the original clip therefore preserving the original audio which we'll relink later in Final Cut. 3. Now open the inspector window so we can do some tweaking. Under "Frame Controls" tab, click the button to the right of "Frame Controls:" to enable this feature. Once enabled, select "On" for "Frame Controls:" Underneath the "Retiming Control" section is where we'll make changes. For "Set Duration to:" click the radio button where it will allow us to put in a duration. NOW THIS IS THE MOST IMPORTANT PART, OPEN THE COPY OF THE CLIP WE CREATED THAT STILL HAS THE 30 FPS TIMEBASE AND PUT IT INTO THE FINAL CUT PRO TIMELINE, DELETE THE VIDEO TRACK AND ENSURE THE TIMEBASE OF THE SEQUENCE IS SET TO 23.98, GO TO THE END OF THE AUDIO AND READ THE TIMECODE, THIS WAY WE'RE ENSURING THAT WE'RE RETIMING (SPEEDING BACK UP) THE CLIP SO IT WILL MATCH THE ORIGINAL AUDIO. This number is the timecode you'll put into the duration ﬁeld, (example01:08:01:06) then depending on how well you want the footage to look also taking into consideration the time it will require to compress it choose the desired quality under "Rate Conversion:". I choose, "Best (High quality motion compesnation)" to get superb results, however this increases the compression time. Important: Make sure that no matter which codec you use that it remains as a 23.98 fps timebase. 4. Now take the treated clip into Final Cut, we've already set up the sequence from step 3. Add the clip to video track 1, relink and presto! I've extensively tested this and ﬁnd it to be the best way to convert 30p to 24p for the time being, or at least until and hopefully Canon comes up with a ﬁrmware to address the issue. Additional Tips: If you have multiple clips to treat, in Cinema Tools, choose "Batch Conform" locate one ﬁle where all the ﬁles are located. Then setup batch conversions in 5dMkII 30P to 24p conversion.

June 25th, 2009, 03:26 PM	#10
Christopher Lovenguth Regular Crew Join Date: Mar 2009 Location: Brooklyn, NY Posts: 85	I'm starting to think this is a FCP issue since I have no problem with sync from H4N (recording 48Hz 24bit) editing in Premiere Pro exporting out using uncompressed audio and ProRes. However I hope someone can tell me why after this being successful, I have audio sync issues just converting this footage to 720p H.264 uncompressed audio every time?