-
Is there a way to encode continuous time onto an audio track?
-
Reading up on this, SMPTE involves storing time on every frame of a video and similarly storing time on every sample (?) on an audio track.
Yes, SMPTE TC identifies each frame with a unique number. No. it does NOT sub-divide into every audio sample. A simple example would be 30 fields per second and 48000 audio samples per second. That would come out to one SMPTE value for each 1600 audio samples. Note that audio for video is essentially universally sampled at 48000 (48K) samples/sec. Virtually all video editing software uses 48K audio sample rate. If you import (for example) 44.1K audio track, it will need to be trans-coded up to 48K, either explicitly by the user, or in many cases, automatically by the NLE software.But, surely, an mpeg video file already has time embedded in every frame, surely? What does SMPTE add?
No, there is no built-in TC. Of course implicitly, if you know the time when the clip started, you can derive the TC for each frame. Note, however that it is rare that a video camera or audio recorder are exactly precise. All gear drifts. Even professional gear that cost as much as your car. But professional operation uses expensive temperature-compensated crystal oscillators for each camera and audio recorder to maintain sync at least over a few hours. Even the most expensive and stable gear is typically “jam-synced” together at least every 4 hours (at meal breaks, for example).And the same with audio. If the starting time is known, then you can derive the time from the starting point and the number of samples. In fact, this is the standard for audio recording. A “WAV” (or “AIFF”) file is a subset of the RIFF standard which stores different kinds of information in identified “chunks”. For example a WAV file contains a block that identifies itself as a RIFF/WAV file, the number of tracks/channels, the number of bits per sample, and the number of samples per second. A super-set of WAV is called BWF (Broadcast Wave Format) and BWF uses extra chunks to store the starting SMPTE TC. But for recorders that do NOT record audio into BWF, a compromise is to simply record the running SMPTE TC onto a dedicated audio track. SMPTE TC was originally designed to be recorded as an audio signal onto one of the audio tracks of an analog video recorder.
I read some web material on SMPTE in audio and it embeds the digital data in the audio, so you need a special player to strip it out, otherwise you can clearly hear it.
Yes. Back before digital video and audio recording, SMPTE TC was recorded as an audio stream on a spare audio track. Obviously, you don’t mix the SMPTE TC track into the audio mix. You only use it to identify the location of the clip in the time-line. Of course, that uses up one of the audio tracks and there is no real need for that if the digital video camcorder or audio recorder handles SMPTE TC internally. But amateur gear is NOT designed to handle SMPTE TC as it is considered a professional feature. So it is included only in higher-price pro gear.Searches on equipment tend to lead to this product every time: https://tascam.com/product/dr-701d/ That does just audio, with masses of features. I don’t mind the price if it actually works with Vegas Pro 13… and preferably earlier versions like MSP11 which I have on a laptop I travel with.
Yes, that is an example of a “pro-sumer” audio recorder that incorporates timecode and writes audio tracks into a BWF format file. Similar “low-price” products are the recent Zoom F4 and F8 recorders. Note, however, that ANY recorder can be used if you have a SMPTE TC generator and a spare audio track.Whether the BWF style, or the dedicated TC track method is compatible with your NLE (non-linear editing) software is something that you will have to research for yourself. I admit I don’t know what “MSP11” means?
I also see a lot of stuff about timecode generators. Why is this a separate function, when everything containing a microprocessor has (or can have) a real time clock for about $1? (I am an electronics hardware/software designer).
Yes, you are correct. Note, however two significant factors:- First, the real-time clock in most microcontrollers is not as accurate as most SMPTE TC gear. Of course, that could be remedied by simply substituting a TXCO (Temperature-compensated Crystal Oscillator) in place of (or in addition to) the microcontroller clock.
- Second, the ability to record the starting TC into the extra chunk in the BWF format is reserved as a “professional” feature for higher-price gear. Most consumer customers don’t need frame-accurate TC . And of course you are correct that adding the TC chunk to audio or video recording is essentially trivial here in the 21st century with microcontrollers everywhere, down to and including toothbrushes, etc.
Presumably the DR-701D contains all that and you just feed in Line In. I have a line level audio signal connection from the aircraft intercom.
Yes, there are many discussions and information on recording from aircraft intercom systems. Note that there is often interference from other aircraft systems. A very common problem is background “whine” from generator/alternator power sources on-board. This is typically ignored by users, and masked by ambient noise in the cabin/cockpit. But is is more annoying to people viewing/listening to the recording after the fact.Note that a common technique is to take a very small “lav” mic and to simply insert it between the headset ear-cup and the user’s ear. That captures the intercom audio without requiring a hard connection into the intercom system. Note also that there are commercial adapter cables which have the special aircraft intercom connector (PJ-055 and PJ-068). One popular source is https://www.mypilotstore.com Users have sometimes (frequently?) resorted to using isolation transformers and/or notch filters to handle the aircraft power noise.
Indeed, it is trivial to program an Arduino (or equivalent) microcontroller to generate SMPTE TC. But the accuracy will be directly dependent on the accuracy of the microcontroller clock. Even the standard RTC (Real Time Clock) solution typically used for microcontrollers (like DS1307 et.al.) is orders of magnitude LESS accurate than typical TXCO clocks used for SMPTE TC generation. But for an application like yours, frame-accuracy is not really necessary, and the less accurate RTC method would easily get you into the ballpark when syncing audio and video tracks in your NLE.
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Yes. I would think that you could do a “clap sync” when you start the camera and establish the audio recorder time code when the camera starts. Then you could start/stop the audio recorder which would continue to count the time code even when not recording.
As I said in the other (concurrent) response, the accuracy of professional SMPTE TC systems is much better than 5PPM because we are talking about frame-accuracy, which is 30x better than simple “wall-clock” time accuracy. And even at that, the cameras and audio recorders are typically jam-synced every 4 hours (or more frequently).
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Thanks Richard for a brilliant clarification of all this.
I just need to check out the Vegas workflow. Some google hits go back 10 years so it looks like it has been in there a long time.
Re aircraft audio, if you connect into the intercom (a.k.a. “audio panel) – this is usually done by connecting into a spare headset socket – you get the benefit of the radio squelch i.e. total silence unless somebody in the cockpit talks, or ATC talks. That is how this one
https://vimeo.com/184011280
was done. Previously I used the mike in the headset method (I bought a tiny mike and a semi-pro preamp from Sound Professionals in the USA) and that is cumbersome as it keeps falling out when you move about, and it does pick up the same aircraft noise which you hear.I have sent this thread to Tascam and maybe they will confirm this should work… it needs to generate its own independent timecode.
I am amazed mp4 video doesn’t store the time in each frame! All the players such as VLC must be faking it, by counting frames. That is something to which I will have no solution, but then I don’t need timing to better than a second or two. However, when I have loaded video and audio from separate sources (I use the $100 Tascam mp3 recorder) onto the timeline, there is zero evident offset after a 7hr flight, so it is good enough despite the camera spending time at -25C, etc.
Some contents or functionalities here are not available due to your cookie preferences!This happens because the functionality/content marked as “Vimeo framework” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.
-
Sorry, above video is the wrong one, but I can’t edit this post. This is the right one
https://vimeo.com/178217220Some contents or functionalities here are not available due to your cookie preferences!This happens because the functionality/content marked as “Vimeo framework” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.
-
Well, you aren’t exactly doing “lip-sync” so you probably don’t notice any significant drift. However, if you were doing something requiring precise picture/sound sync, you could expect several frames of drift after even 10 minutes at room temperature.
HOWEVER as even inexpensive, mass-produced consumer gear inevitably gets better from more accurate crystal manufacture, that is going up to 20 or 30 minutes with less than 1 frame of drift.
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Searches on equipment tend to lead to this product every time: https://tascam.com/product/dr-701d/ That does just audio, with masses of features. I don’t mind the price if it actually works with Vegas Pro 13… and preferably earlier versions like MSP11 which I have on a laptop I travel with.
I just minutes ago read this characterization in another forum:
“And the DR-701D TC drifts like a ship with out a rudder. So except when it’s constantly jammed with HDMI TC it is not going to stay in sync.”
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Interesting… it must depend on what they are doing with it. It can’t possibly be bad enough for what I want : )
The DR701 also has a terrible battery life. External power (USB, or their battery pack) is really necessary for most things. OTOH the little $100 recorders can do 50-100hrs of recording on 2xAA.
I can see some people, if mono is OK, will be recording a timecode on one of the two audio channels, from a timecode generator. Then you can use any cheap sound recorder.
There is a lot of stuff online on this. One could read it for days.
-
I have one more Q if I may:
Does video have no timecode in it at all? Not even on the first frame?
Because if it doesn’t, there is little point in timecoding the audio. One needs to timecode the video also.
But I am sure I am wrong, because the HDMI-derived timecoding of audio which this does
https://tascam.com/product/dr-701d/overview/
could not work. There must be a timecode coming out of the HDMI data. Doesn’t this timecode also get recorded in the mp4 file, in the camera? -
Does video have no timecode in it at all? Not even on the first frame?
Professional cameras do. But it doesn’t sound like that is what you are using.
Note that if you can set the time/date on the camera, and the video file has a proper time/date stamp, that gives you a rough time reference when the video started (assuming you set the camera properly).There must be a timecode coming out of the HDMI data.
No. That Tascam page is talking about “genlock” (locking the RATE of the video camera frame clock with the rate of the audio recorder sample clock. That is a different matter than timecode which is the actual hour minute second frame serial number of each frame. Here is part of a tutorial from B&H:Editors should be aware of a few issues when transferring video to another deck, device, or NLE computer system. The time code will not transfer if you connect via analog video, composite RCA, S-video (Y/C), component, or HDMI interfaces. Time code is carried through iEEE1394 (DV), RS-422 protocol, or through a camcorder that incorporates a separate time-code channel, typically with a BNC connector. Time code is crucial, especially if you want to do a “batch capture” with your NLE. Batch capturing is the process of capturing selected clips from a recorded-on-source tape, based on a log or shot list created by the user.
https://www.bhphotovideo.com/find/newsLetter/Time-Code.jsp
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Timecode and Genlock are two different and independent functions, very frequently used together, and often confused.
TIMECODE is a method of indexing each frame of video (and a frame’s worth “block” of samples of audio). The timecode includes hours, minutes, seconds, and frame numbers. This MAY be the actual time-of-day (“wall clock” time), or it may be some other arbitrary number as decided by the production workflow.
GENLOCK is a method of “locking” the frame-rate of the camera with other cameras and/or audio recorders. It cancels “drift” where unlocked gear runs at slightly different speeds because of normal variations in manufacturing tolerances. It sets one “master” reference which everyone else agrees to use (instead of their own internal clock). The “master” may not be any more accurate than each piece of gear, but the point is that they all run at the same RATE.
Back in the days of analog video, all studio cameras were required to be GENLOCKed together so that they could be seamlessly switched/dissolved etc. in the production switcher. Modern digital gear tends to use “frame-buffering” where digital video coming from unlocked cameras is artificially “synced” together. This DOES cause a slight latency, and for that reason some higher-end digital gear still supports genlock.
Note that modern designs have started to exploit GPS where BOTH the time (TIMECODE) and the rate (GENLOCK) can be locked together by using the wireless signals from the GPS constellation of 32 satellites. (Or the alternatives like the Russian GLONASS or the proposed systems from the EU (Gallileo), China, Japan and India.)
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder.
Reply to this Discussion! Login or Sign Up