Caching Sound

Still doing some test, and there is something makes me crazy.

I add a Video, Cut (I tried both, K and Shift+K) the first 20 seconds for example, play, and sounds plays from the Start as if I don’t cuted the video, and doesn’t matters where I’m telling to start, always starts as the video recently added. After some combinations of moving the green vertical bar (I don’t know the name sorry), playing with Play button, or Control+A and so on, sometimes, sound looks like start in the right place, but again, same happens.

I tried caching the audio, and seems this solves the problem, but the video then plays a lot slower (Blender taking only 300MB of RAM), and I can’t check anything if it’s right/synced etc… I tried to Pack Audio, but then Blender crashes.

Some idea?

I tried caching the audio, and seems this solves the problem, but the video then plays a lot slower (Blender taking only 300MB of RAM), and I can’t check anything if it’s right/synced etc…

Have you tried selecting ‘AV-Sync’ in the timeline Playback menu

Yes, I tried changing to different values.

What OS and Blender version? 32 or 64bit? And what video / audio codecs?

Caching sound should improve things, it uses more memory because it’s putting the sound into memory in order to stand more chance playing back in sync.

Have you tried hiding your movie track and just playing the audio, does that playback smoothly? It may be the video strip decoding that is the issue and not the sound decoding.

Windows XP SP3 32bits. Video is in MPEG2. Blender Version, I tried with 2.56a, and 2.57RC1. Thew machine is a P4 3GHZ with 2GHZ RAM

Caching sound should improve things, it uses more memory because it’s putting the sound into memory in order to stand more chance playing back in sync.
Yes, I understand this, but with 300MB is not much for a playback so slowly.

Have you tried hiding your movie track and just playing the audio, does that playback smoothly?
It may be the video strip decoding that is the issue and not the sound decoding.
Sound is smoothly, slow is the video playback, and plays from the start of the video, althoug I deleted a part.

The audio is muxed (built in) with the vision. The mpeg format relies on keyframes and predictive frames, so not all frames exist. This makes it hard for blender to decode if you cut the file at some point that does not have keyframe (or real frame). So the audio tends to drift out of sync unless played from the start. In 2.49 you could mixdown audio to a wave file, effectively replacing source audio, but you really need to convert the mpg2 files to an edit friendly format.

Ok, I’ve found 64bit builds to perform slower than 32bit on Windows.

Sound is smoothly, slow is the video playback, and plays from the start of the video, althoug I deleted a part.

If sound is smooth then look to solving video playback, does video alone run smoothly with AV-Sync activated and audio track muted?

So as 3pointedits description and suggestions generate proxies or convert MPEG2. My preference is always proxies.

Audio mixdown is possible in 2.5x just not so visible, render video but choose .wav as output not video formats, it renders the movie to the preview window but only exports the audio. :wink:

I don’t have problem with this. What could be a good format to manipulate it? DV+PCM?

Proxies? Uhm…I don’t know what is this sorry :S I will try to check the Blender Wiki to check

Proxy generation is not currently possible with 2.5x (Devs would rather find alternative solution than reimplement I think) however 2.5x will use proxies if they are available. ie: created outside of Blender 2.5.

It may not be worth the effort for your purposes but the blog in my sig has proxy generation scripts. However there are many ways to convert your video if you choose that route that are far less involved than producing proxies but don’t necessarily provide a decent transcode with regard to so called ‘quality’. :slight_smile: