When using AMT to transcode (or re-wrap) we use the WriteAmtTrackSample method and pass in buffers of audio and video data. A change in AMT 2.11 was that the audio buffers needed to be equivalent to a frame of video in duration otherwise errors in the audio MXF files to do with how the duration is described crop up. Thus with that issue identified we have modified our code.

However, can you provide guidance over what buffer sizes to use with fractional frame rates such as 23.976, 29.97 and 59.94 as these do not divide into 48000?

For example with 23.976 fps we are using an audio chunk sample size of 2002 (48000/23.976) but we are concerned that over long period an eventual difference between the number of audio and video samples might occur. Is it better to round up or down?

e.g. with 59.94 you get 800.800800800800 so is it best to use 801 or 800?

Thanks

Nick

Nick,

This is a good question. I'll check with our development team...

- Rob

Nick,

SMPTE standards actually defines how to handle this for NTSC frame rates via a so-called 'cadence' of integer sample counts. For 29.97 fps and 48 kHz sample rate, you are supposed to use 1602, 1601, 1602, 1601, 1602 samples and then start over again. The details are not Avid specific and defined in the respective standards. For a brief overview, try a google search like 'audio sample count 1601 1602'. Also, the AMT sample application has a basic implementation of this.

Can you confirm if the audio smaple pattern for 23.976 shoudl be 2002 for all frames oe whether it has a 499 x 200 and then a 2003?

Thanks,