Syncing issues ADPCM to WAV

A

andrew00

Well-known member
Hi,

I'm not very technical at this whole audio thing, so I'll try to be as clear as I can, but forgive my stupidity.

I've been recording some audio through my lav mics (ATR35S's) into an mp3 player - Transcend t.sonic 630, which records to wav files that are ADPCM.

I've then been bringing them into Soundforce and saving as WAV files (I guess normal PCM wav files) and then importing them into Sony Vegas for syncing, often with video recorded at the same time.

However, as I try to sync these audio sources up (say two, a conversation, with one mic/mp3 player per person) I've been finding I can't get the syncing right.

What happens is that if I sync these files at the start, then by the end of the recording, even if it's as short as 10 minutes or so, they have gone out of sync. Likewise if I sync at a point at the end, they are out at the start.

Please would you let me know what I'm not doing right here, as I don't quite understand how two files, both recorded in the same wav and converted in the same program can get out of sync with one another.

Likewise I'd therefore appreciate help on how best to solve this problem so the files can remain in sync.

Thank you kindly!

A
 
andrew00 said:
Hi,

I'm not very technical at this whole audio thing, so I'll try to be as clear as I can, but forgive my stupidity.

I've been recording some audio through my lav mics (ATR35S's) into an mp3 player - Transcend t.sonic 630, which records to wav files that are ADPCM.

I've then been bringing them into Soundforce and saving as WAV files (I guess normal PCM wav files) and then importing them into Sony Vegas for syncing, often with video recorded at the same time.

However, as I try to sync these audio sources up (say two, a conversation, with one mic/mp3 player per person) I've been finding I can't get the syncing right.

What happens is that if I sync these files at the start, then by the end of the recording, even if it's as short as 10 minutes or so, they have gone out of sync. Likewise if I sync at a point at the end, they are out at the start.

Please would you let me know what I'm not doing right here, as I don't quite understand how two files, both recorded in the same wav and converted in the same program can get out of sync with one another.

Likewise I'd therefore appreciate help on how best to solve this problem so the files can remain in sync.

Thank you kindly!

A
Click to expand...

For the sound recorded in two devices to stay in sync over the duration of a shot, the sample clocks in both devices must run at exactly the same rate. If they aren't, even though the files were recorded simultaneously, when placed side by side on a single playback device such as when you import them into Vegas they will be slightly different lengths. The sample clocks determine the number of samples per second that are recorded - if the clocks in your recorders are slightly different, each second of audio will have two different total sample counts. When played back together, one file will play slightly faster or slower than the other. Getting sample clocks to run exactly on spec takes careful engineering, high quality components, and intense quality control in manufacturing, something that manufacturers of consumer gear just can't afford to do without raising the price of their product far above what you'd want to pay. If the music MP3 you bought from iTunes plays 0.1% faster on one player than it does on the other you'll never notice but if you're trying to record two files that stay in sync over 10 minutes that amount of difference will be very noticable indeed. Even professional recorders costing several thousands of dollars will eventually drift off sync unless you take special steps to slave their clocks to a common source.
 
Last edited:
Ah, I see, I wasn't aware of that, how confusing it all is!

So, even if I'm using two identical mp3 players they can still go out of sync.

Great!

Is there anything one can do about it logistically, I can't afford anything expensive so really have to either make do with what I've got or just dump them - can I use a software fix or something or do I just have to keep cutting them up and re-aligning them when they go out?
 
Is it possible for you to just use one mic/mp3 player for the shot? If I understand correctly, that should work. Other than that, you could probably go in and do some surgery on the sound, cut up each of the dialog and place them in the correct spots and then maybe use just one of the tracks ambience noise for background. Unless the ambience noise is something complex like a diner environment. Just a suggestion.
 
Well I'm filming bands, so it's not really viable to use one mic. What I do is mic up each band member and have them play freely, the idea being they are free to roam.

Alas if I can't get them in sync it will kinda ruin the song.
 
And, thinking about it, if this is such an endemic problem then how do people without this super high end equipment record audio to keep it in sync - do they have to micro manage ever single line to make sure it's perfect?

As naturally my intention is to film many bands and with many instruments if one is out of time then I'm screwed - esp as the cheapo mics I'm using (ATR35S's) can, if placed inaccurately, have a lot of bleed from the other musicians - I don't want a situation where another musicians playing is being picked up, and therefore the out of sync is v. pronounced b/c you can hear said musician playing both on his own mic and also on another band members.
 
One way it's done is to shoot the band to playback as they mimic actually playing the instruments.

It doesn't take super high-end equipment to stay in sync but it may take several thousand dollars worth of proper professional gear. "High-end" is a relative term - do you define that as a $1000 Tascam, a $4000 Sound Devices 4-channel, the new $6000 Sound Devices 8-channel, or a Deva or Cantar costing in the range of $15000 to $18000?
 
Last edited:
Thanks for the replies so far, I appreciate your time helping out a clueless person like me!

Let me explain the concept a bit more because budget isn't high at all, we're talking a couple hundred max.

My goal is to take a band, say of 5 members, and film them playing one of their songs live in a non-typical environment in one continuous shot. I could do this with a sound guy and a boom, but the objective is to let the musicians move freely however they want around the environment and not be constrained by such typical logistics like keeping in mic range etc. In essence, to create an 'anti' music video, where nothing is planned, everything is spontaneous, and the camera records what's happening as best as it can.

There's a couple people doing something similar already, such as these:
1 - http://www.blogotheque.net/-Concerts-a-emporter- (they originally just used a single mic on boom or a zoom h4, but with success have now moved to a full wireless system I think, e.g. the vids with REM etc)
2 - http://www.blackcabsessions.com/sessions.php (these guys are from the uk, but their concept is filming in a cab and so don't have freedom of movement like I want)

In essence I'm trying to go the other way and have ultra freedom of movement for all musicians. I've done one or two tests with some friends that you can see here:

http://www.dailymotion.com/andrew00

They're not perfect or anything, but I think you can see the idea a bit, although there isn't too much movement as my mates were all nervous! The equip used here was 1x ATR35S mic into 1x Transcend t.sonic 630 per person. I also had a shotgun on my camera that I used to give a more natural feel, but that feed was v low.

My current equipment is:

- Camera (Dvx, hopefully soon Hvx)
- SM57 mic, Rode RT2 shotgun mic
- Boom & **** load of leads
- Zoom H4

(this was my equip I had before this project, but now I also have)

5x Audio Technica ATR35S wired lav mics
5x Transcend t.sonic 630 mp3 players

I did a lot of research into what to record on and found that a wireless system would be well beyond my budget, but since my aim was freedom of movement I was determined to have 1x mic per musician. Ergo I needed to record the mics, hence the purchase of the Transcend mp3 players, which record decent quality (not exception, but good enough for web usage) audio in this ADPCM format.

I wanted to use iRiver iFP mp3 players, but it seems the podcasters are snapping them up with haste.

Is there any way you lovely gentlemen whose time I greatly appreciate in reading and answering!, might get around this syncing problem.

As it stands I am happy both with the ATR35S's and the Transcend mp3 players. However if I cannot sync the audio files from those sources then I'm screwed and will have to a) abandon the freedom of movement concept until I can afford a full wireless system or b) stick to recording with the shotgun mic/boom/h4 system, and so miss lots of audio (and need a seperate sound guy) until a) is viable.

Naturally I want this system to work so I'm keen to find a solution rather than compromise.

As it stands, therefore, is is a) the inherent limitation of the equipment (the mp3 players) that is getting in the way, and ergo there is NO WAY to fix the sync issues w/ software...or...is it a matter of taking the raw ADPCM wav files and applying some software thingamajig to get them all converted in the same way and keep them in sync?

Thanks, I know this is a big post but I'm really passionate about this project and, well, I'm passionate enough to be a bit pushy in my desire for a solution!
 
It's fiddly but you might try using an audio workstation that lets you increase or decrease the length of a clip without changing the pitch. When you shoot, use a conventional clapper board slate and be sure to record one at the start of the clip, another at the end. Drop your video and all the audio tracks into the workstation and line up the head slates. Scroll to the end of the tack and find the tail slates. Stretch or compress each track until all the tail slates line up with the marker in the video track.
 
OK, mostly "what he said". But the drift on digital systems of pro-sumer and above is quite small, almost nonexistent compared to the analog systems they replaced. So with a single recorder you generally don't have much of an issue. However with multiple recorders of consumer quality you could have considerable drift. And your camera is going to have drift (very small but...). Now to complicate things your doing a number of conversions each of which could be causing you problems. I would bet that your MP3 recorder records at 44.1, and of course if your going to DV it needs to be at 48 so one of those steps is doing that conversion and poss adding a drift. And lastly how are you judging sync? You have five non isolated mics so your going to get a jumble of out of sync sound no mater what you do. Sound travels at, well the speed of sound, 344 miles per second. Fast but you will still have a delay that is audible (probably) with sounds from other band members hitting the mic. And this will be true for all the mics. Also your camera is seeing at the speed of light, 186,282.397 miles per second. A LOT faster. So even with everything running perfectly you have sync issues that need to be fixed.

Back to how do pro's do it.
Better recorders, word clock locked if your using more than one, and word clock or black burst locked to the camera. Then hand syncing in post to compensate for some of the above variables and in general shorter takes.

What you might try doing is slating head and tail with a proper slate. Then in post sync up the pops. If your pops don't line up take the take that fits the video best, time stretch whatever to get it to fit from start to finish. Then stretch/ shrink the other tracks so the pops match. It is also poss that the recorders can't keep up with the data flow of a ten minute uncompressed recording and are "dropping frames" (taking little pauses to catch up). You might not even be able to hear the gaps if they are really short. A pro recorder would be out of business if they did that but consumer MP3 recorders customers would probably never notice. Unless they tried to use them to record sync sound. Hopefully this isn't the case because there isn't a good workaround if this is what is happening.
 
andrew00 said:
Thanks for the replies so far, I appreciate your time helping out a clueless person like me!

...
Thanks, I know this is a big post but I'm really passionate about this project and, well, I'm passionate enough to be a bit pushy in my desire for a solution!
Click to expand...

I like the idea - I think music in the environment is an interesting concept. But to do it justice you need to do it right and that just might not be possible on a shoestring budget. I don't mean offense but you've heard the saying that you can't make a silk purse out of a sow's ear. If the pro's could get good results with bargain basement gear that's what they'd use - most big-time soundies buy their own gear packages and every dollar spent on a recorder or mixer is a dollar not available for the mortgage, vacation, or the kid's college fund. But to get professional quality results there's a certain minimum price of admission that you just have to bite the bullet and come up with.
 
Thanks for the replies, lots of interesting info I greatly appreciate.

I appreciate what you're saying Steve, I'd love to be able to afford the top notch equipment, and hopefully one day I will be able to. The thing is, I'm nowhere near it at the moment, so it's either a case of wait unti the day I can afford it, or simply make do until that day. It's not perfect, certainly, but if it's a case of producing something decent but not exceptional now versus something fantastic in five years...well, I'm impatient! That is, however, one of the reasons I'm really grateful to all the advice I'm getting here, because at worst, if I'm doomed to not-great audio, then hopefully I can at least make the best out of the sow's ear, at least in short-term before I can get that lovely silk!

Thanks also chaps for the drifting advice, I guess you'd say I was your typical consumer in that I had no idea such a thing existed! I also appreciate the inherent and perhaps unavoidable issues that might arise with my somewhat less than superb equipment.


The way I'm seeing my workflow for future projects is as follows, please let me know if I'm incorrectly viewing this:

1 - Mic each musician up and get their recorders recording. Then, get everyone together and create a head (start) sync point, essentially by getting all the musicans together so the mics are close and shouting something to mark the point! When finished, repeat this process to create a tail (end) sync point.
2 - Import each ADPCM mono wav file into Soundforce. Select 'save as' and save as 16 bit 48 khz. Repeat for the wav file from each mic.
3 - Import these audio files onto my Vegas timeline and trim each clip so the clip starts at the head and ends at the tail points.
4 - Line these up as best possible in order to sync as best possible.
5 - Using the time stretch feature, drag the files so that they are all in line with the master audio track (my camera) both in head and tail. (This should be preferably done with a visual cue.)
6 - This should ensure that, at best, each clip begins (head) and ends (tail) at the exact same point, and should theoretically be in sync (unless is goes out of sync at random points during the video, but in that case you're screwed anyway!).

That's what I'm reading, please let me know if my feeble mind is up to speed and, thanks again, I and my feeble mind appreciates it!

a
 
Understanding that traditionally, as Steve said, you use playback to record a music video, what you want to do is record a live, roving, concert. Giving your budget you could easily rent 5 w/l for the day of shooting and record into a multitrack recorder. Include a live boom track for the mix and you should be able to record the whole thing for under $500. Of course you could just hire a pro to do it for about the same $ and not have to sweat audio anymore.

Cheers
 
andrew00 said:
...
1 - Mic each musician up and get their recorders recording. Then, get everyone together and create a head (start) sync point, essentially by getting all the musicans together so the mics are close and shouting something to mark the point! When finished, repeat this process to create a tail (end) sync point.
2 - Import each ADPCM mono wav file into Soundforce. Select 'save as' and save as 16 bit 48 khz. Repeat for the wav file from each mic.
3 - Import these audio files onto my Vegas timeline and trim each clip so the clip starts at the head and ends at the tail points.
4 - Line these up as best possible in order to sync as best possible.
5 - Using the time stretch feature, drag the files so that they are all in line with the master audio track (my camera) both in head and tail. (This should be preferably done with a visual cue.)
6 - This should ensure that, at best, each clip begins (head) and ends (tail) at the exact same point, and should theoretically be in sync (unless is goes out of sync at random points during the video, but in that case you're screwed anyway!).

...
a
Click to expand...

That's pretty close, except a shout doesn't make the best slate. The old fashioned Hollywood clapper is much better because you have an exact visual mark on the tape as the sitcks come together plus a very distinctive peak in the audio waveform that you can easily see to use to align - a shout's waveform is more like a 'smear' while the clap is a spike.

I looked up the web references to your mp3 recorder and it doesn't tell much about them except they're recording at best at 44kHz. You might try playing your audio and re-recording it into Vegas rather than converting the file and importing it digitally. As was discussed before, some timing errors are caused by recording a file on a device with one clock speed and playing it back on another device with a slightly different clock speed. When you do that it plays beack either slightly longer or slightly shorter than when originally recorded. If you play it back on the device that recorded it there's less chance the length will change. It won't be perfect because of temperature differences between recording and playback but it'll give you a fighting chance.
 
Re - the clock, am I understand this right. Every device has a clock within it that counts the time, 1 second, 2 seconds etc. However at some point, like down to the micro micro seconds or something, these clocks might be not quite accurate, i.e. rather than 1.0000 seconds per second they might record 1.0001 or something, ensuring that over a long period of continuous recording (like I might do) they are prone to going out of sync and out of perfect time keeping aka - drift. This isn't a problem really with one audio recording device, or when you have a master clock all devices are taking their time from, but it is a problem when you are doing what I want to do when recording multiple sources independent of one another, which might all simultaneously have different clocks (esp pronounced since they aren't exactly top of the range equip).

Re - Recording - Right, I think I understand, how would be the best way to re-record it into Vegas? Would it be to play the song on the player and plug into the pc through the line in port and record directly that way? Or are we talking more plugging the headphones slot into the PC and recording that way, or is that the same thing? (I mention both because the player doesn't have a direct line out port and I don't think the line in/mic in port is also used as a line out).

Or did you mean plug it into the pc via usb, play the recorded file in VLC media player or whatever, and also use audio recording software on the pc to record the playing file that way? That way I'm recording in the audio recorder at 48khz the signal being played, and as such this clock drift is minimised because we're recording all the different sound files again through one source - the software, which will have it's own identical clock, ensuring that the audio signals themselves have the same clock?

This would in turn reduce the potential negative effects of 'saving the original adpcm source files as' 48khz sound files, a process that might introduce additional drift (and the process I was hitherto taking in soundforge 9). Or can this be minimised by setting Soundforge as the master clock source when 'saving as', which I don't know how to do, or am I reading this wrongly?

Re - the recorder: I don't know the specifics, but I know the best recording mode is indeed 44khz and is meant to record for 6/12hours (stereo/mono, mono from external mic), so I don't know if that affects how we can judge the quality - I'm happy to record some samples if you think that's help? (online manual here - http://www.transcendusa.com/Support/DLCenter/Manual/TS1GMP630_EN.pdf )

The manual also says on page 54 the following:

• Recording Format: ADPCM (WAV)
• Compression Rates: 32Kbps to 320Kbps
• Erase Cycles: ≥ 100,000 times
• Signal-to-Noise Ratio: ≥ 90dB

I don't know if that helps judging the quality/drift, but it might!

Thanks again!
 
Last edited:
I didn't catch this before but ADPCM is a compressed codec like MP3. Even if your recorders had some kind of clock in they wouldn't lock together probably. This is really not a good choice for sync recording.
I'm sorry I can't help much with software since I've been Mac based for quite a few years now. I have used Sound Forge, but it was like version 2 before Sony bought it.
Anyway your sort of working with todays equivalent of a bunch of cheap portable cassette recorders. I would forget trying to make them work "correctly" they just aren't going to. If your lucky they will run consistently. I would just accept the fact that what you get from them is a raw un sync set of recordings and manually sync them up. I don't know if Vegas is really the right application for this. My impression was that it was primarily a video editing application. For the kind of surgery you will probably need something more along the lines of Pro Tools would be a better choice. And actually I would just cut you video with the camera audio as a guide and then find a sound person locally to help with the sync of all the feeds. I suspect that not only do they "drift" (and yes you got that right, internal clocks running very slightly slow or fast) but that they are also not 100% consistent in there recording. They have a lot of processing going on to record ADPCM files and I bet they waver depending on the workload.
 
andrew00 said:
Re - the clock, am I understand this right. Every device has a clock within it that counts the time, 1 second, 2 seconds etc. However at some point, like down to the micro micro seconds or something, these clocks might be not quite accurate, i.e. rather than 1.0000 seconds per second they might record 1.0001 or something, ensuring that over a long period of continuous recording (like I might do) they are prone to going out of sync and out of perfect time keeping aka - drift. This isn't a problem really with one audio recording device, or when you have a master clock all devices are taking their time from, but it is a problem when you are doing what I want to do when recording multiple sources independent of one another, which might all simultaneously have different clocks (esp pronounced since they aren't exactly top of the range equip).

Re - Recording - Right, I think I understand, how would be the best way to re-record it into Vegas? Would it be to play the song on the player and plug into the pc through the line in port and record directly that way? Or are we talking more plugging the headphones slot into the PC and recording that way, or is that the same thing? (I mention both because the player doesn't have a direct line out port and I don't think the line in/mic in port is also used as a line out).

Or did you mean plug it into the pc via usb, play the recorded file in VLC media player or whatever, and also use audio recording software on the pc to record the playing file that way? That way I'm recording in the audio recorder at 48khz the signal being played, and as such this clock drift is minimised because we're recording all the different sound files again through one source - the software, which will have it's own identical clock, ensuring that the audio signals themselves have the same clock?

This would in turn reduce the potential negative effects of 'saving the original adpcm source files as' 48khz sound files, a process that might introduce additional drift (and the process I was hitherto taking in soundforge 9). Or can this be minimised by setting Soundforge as the master clock source when 'saving as', which I don't know how to do, or am I reading this wrongly?

Re - the recorder: I don't know the specifics, but I know the best recording mode is indeed 44khz and is meant to record for 6/12hours (stereo/mono, mono from external mic), so I don't know if that affects how we can judge the quality - I'm happy to record some samples if you think that's help? (online manual here - http://www.transcendusa.com/Support/DLCenter/Manual/TS1GMP630_EN.pdf )

The manual also says on page 54 the following:

• Recording Format: ADPCM (WAV)
• Compression Rates: 32Kbps to 320Kbps
• Erase Cycles: ≥ 100,000 times
• Signal-to-Noise Ratio: ≥ 90dB

I don't know if that helps judging the quality/drift, but it might!

Thanks again!
Click to expand...

No, I meant play it out through the analog outputs of the recorders, feed the signal into an audio interface and re-digitize it as you record it onto your PC. Not normally a good approach, but at least it gives you a fighting chance with the timing issues since you're playing it back with the device that originally recorded it, and then re-recording it with the device where you'll be syncing and editing. The chain of events that a digital transfer has to follow - uncompressing from ADPCM followed by a sample rate conversion to get it to video's 48kHz, along with all the other variables, makes it an uphill battle.

Are you using Vegas Pro or Vegas Studio? Which release? Which version of Soundforge do you have.

Luckily you really haven't spent much money on this yet. Before getting too far into using what are basically toys (I'm sorry to sound harsh but we've gotta face facts here - that's what they are) maybe it would be a good idea to step back and re-evaluate the entire approach. If it's a good idea to do the film at all, it's a better idea to make sure it's done right. Your audience only cares about what they see on the screen and won't give you any points for the creative use of bargain basement tools - all they'll care about is the end result. If these devices won't cut the mustard, better to cut your losses now and start fresh with a toolkit that works.
 
You must log in or register to reply here.
Back
Top