Audio Distortion using Frameserver

PJH wrote on 8/17/2014, 11:53 AM
Hi everyone,

I'm in the process of rendering a video from Sony Vegas to TMPGEnc and I've noticed a slight audio "distortion" on the rendered mpg file.

I tried setting the Frameserver setting to "use PCM audio" but it's made no difference.

The audio on the timeline is clean but it's only on the final file.

Does anyone have any suggetions?

I've tried rendering HD (m2ts files) or just standard DV files to mpeg 2 and the result is the same.

Cheers,

Peter.

Comments

john_dennis wrote on 8/17/2014, 12:42 PM
1) Is the audio clipping in Vegas Pro Master Bus Meter?

2) Have you tried rendering to .wav from Vegas Pro, putting it back on the timeline to see if the distortion could be happening before the frame server?

3) Any effects that might cause distortion or clipping?

Sorry for being redundant, but it's a lot easier to shoot the right bug.
PJH wrote on 8/17/2014, 1:08 PM
Thanks, John.

The audio is perfectly clean in the avi file or the AVCHD file. I'm not using any effects on the audio either.

The distortion doesn't sound like clipping. It's more a kind of compression artifact thats evident mostly on the high frequencies.

I've just done a test whereby I made a DV file and then rendered that seperately in TMPGEnc and that audio is fine.

It's just when using the Frameserver.

Cheers.
farss wrote on 8/17/2014, 1:50 PM
Is your project's audio resampling set to Best?

Bob.
flyingski wrote on 8/17/2014, 1:54 PM
In TMPGEnc are you rendering the audio to PCM or Dolby? When making a BD file with Dolby it will lower the volume, whether it's from frameserve or a regular file. To my ears the Dolby encoder flattens the audio as well but I hear no distortion.

I ran the same file through frameserve with it's PCM signpost checked and unchecked and could hear no difference in the resulting TMPGEnc encoded file. They all sounded and looked the same as the file I rendered directly in TMPGEnc.

In short, I couldn't duplicate your problem. Sorry.
johnmeyer wrote on 8/17/2014, 2:31 PM
Two words: project properties.

Bob ("farss") is right to ask about audio resampling in the project properties, but the complete answer is that when frameserving, you have to look at ALL the Project Properties settings and make sure they match what you want for your final output.

The longer answer is this: for regular rendering, when you are using a Vegas codec, the project property settings are ignored, except for the resampling settings (I never understood why these settings are included in project properties, but that's a topic for another time).

However ...

When using Satish's frameserver, that method of exporting video doesn't do any rendering (that's one of the main reasons for using it!), so what video and audio settings are actually going to be used?? There are only two logical choices: either output each video clip using its source properties, or else conform all video on all tracks to some single standard. The former approach doesn't really make much sense because Vegas can combine video from different resolutions, PAR, frame speed, etc., so what would it put out at each point in the timeline?

Instead, it changes everything to fit the project settings.

You can easily and quickly verify what is going on with your video by doing this test: frameserve out of Vegas using the RGB24 setting. Open the resulting AVI signpost in VirtualDub (it's a free 32-bit application that is easily installed). Then, open the VirtualDub File --> Information dialog.

I just did this for some AVCHD 60p 1920x1080 video from my Sony CX700V. I set the Vegas project properties to NTSC DV Widescreen (720x480, PAR 1.2121), and then purposely degraded the project properties audio to 8,000 Hz. I frameserved this in RGB24 and opened it in VirtualDub. Here is the result:



Note that the audio is 8,000 Hz, NOT the original 48,000 Hz.

So, if you actually want to change the sampling rate, definitely follow Bob's advice. However, an even better approach is the set the sampling rate to match your source audio, and then the audio will not be touched at all.