Soundsoap for Edius question

[Bart]

Is the delay of the sound caused by the Soundsoap-filter supplied with Edius 4 a Soundsoap or a Edius bug?

[Zorro]

Can you please be more specific describing your problem?

[SRsupport]

This problem is there along time.(known issue).
Use sound soap in an audio app to clean the sound. Or render the audio out and sync it at frame level.

[Bart]

This is indeed a know issue. The question is: should Grass Valley / Canopus solve this or Soundsoap?

The result of reporting this as a bug on this forum = 0, not even an explanation on why ... I can only guess that Canopus considers this as a Soundsoap bug. But is this so?

Related question: are there affordable alternatifs?

[THoff]

If memory serves me correctly, the problem is caused by SoundSoap needing 1024 (I think) audio samples before it can output anything back to Edius to play. This causes the audio and video to get out of sync.

There is a simple workaround. Export a WAV file of the unprocessed, problematic audio from Edius, and use the standalone version of SoundSoap to clean it up, then use the clean version to replace the audio track in your project.

Where do you get the standalone version of SoundSoap? You get it free when you download an update for the SoundSoap plugin from the BIAS website -- the update includes both. The current version of SoundSoap is V2.1.1 and adds Vista compatibility; there were several interim bugfix releases between the one that comes with Edius and V2.1.1.

http://www.bias-inc.com/downloads/updates

[GrassValley_MD]

This is indeed a know issue. The question is: should Grass Valley / Canopus solve this or Soundsoap?

The result of reporting this as a bug on this forum = 0, not even an explanation on why ... I can only guess that Canopus considers this as a Soundsoap bug. But is this so?

Related question: are there affordable alternatifs?

I do not know the answer to this but I have used several other VST plugins with no problem. Look on the web. You will find many of them out there.


Mike

[7hbg]

...If memory serves me correctly, the problem is caused by SoundSoap needing 1024 (I think) audio samples ...[/URL]

Almost, but not quiet…

This has been explained in the old forum, but since all good knowledge is gone let me try one more time:

Soundsoap, in order to eliminate any unwanted ‘frequency’ components from the sound file needs to convert the sound from ‘time’ domain into ‘frequency’ domain, remove the unwanted frequency elements and then convert it back to ‘time’ domain (samples). The technique to do all that is either ‘Discrete Fourier Transform’ or ‘Fast Fourier Transform’. Both methods need a power of two samples (ie: 512, 1024, 2048, 4096, etc) in order to do the job. Higher the number of samples, more precise (smooth) the output will be. For example, at 48KHz with 512 samples one can expect almost 100Hz resolution (meaning that the smallest adjustment you can make will be 100Hz wide), but with 4096 samples the smallest adjustment will only affect 12Hz (smaller number better). Imagine you have choice of ‘Graphic Equalizer’ with 256 sliders vs. one with 2048 sliders, if you need to be very precise you will choose the one with 2048 sliders. And in essence that what Soundsoap is, it is a ‘Graphic Equalizer’ that is capable of learning the ‘noise’ pattern and using the DSP technique subtract the noise from the sound file without (or as little as passable) effecting any other sounds. That's why the need for 4096 samples before any output can be done.

Problem with Edius vs. Soundsoap:
The way EDIUS serves the audio data for processing to Soundsoap is the way it has it available from the video file: 1 frame at a time (1602 samples @ 29.97 fps, or 1920 samples @ 25 fps). Clearly it is not enough for Soundsoap to do the job. It will take 3 frames to fill up the required buffer (4096 samples). Once the samples are collected the Soundsoap does the ‘magic’. Normally after processing the samples, the processed data would be placed back where it came from, but in EDIUS there is no way of going back (once frame is processed there is no more access to it) the data is placed as soon as it is available (hence the slide).

At the end, no one will admit to a bug. This is just the way things work.

NOTE: All of this information is based on EDIUS SDK ver. 1.8 previously available on Canopus download page. Since it is no longer available to the public (only by invitation) and it is not applicable to version 4, thing may have changed.

[7hbg]

I do not know the answer to this but I have used several other VST plugins with no problem. Look on the web. You will find many of them out there.

Mike

Here is a small test you can do to see the problem:

Place one copy of this wave file on 2VA (it is a short burst of a tone).
Then place a second copy of the same wave file on 1AV, directly underneath the first one so they line up.
Next, play the timeline. You will hear one burst of tone.

For the next test, place SoundSoup plug-in on the bottom wave file (1AV)(all defaults).
Now play the time line. You will hear two bursts of tone. The original and the second burst (the processed data) exactly 4 frames later on NTSC and 3.5 frames on PAL.

Next (on NTSC only), move the 2VA file (the one without plugin) 4 frames to the right and play the timeline again. You will only hear one burst.

[Bart]

Rob, thank you very much for the time you took to explain this.
At least I understand the origin of the problem.