Volume Normalization on Video Playback on Premiere

[ddetton]

I created my own personal MTV using a 1U Premiere chassis and a Crestron control system. It's pretty simple. i created a script for each song on a concert dvd. At the end of each script, I send a command to the control system to grab the next random song. What I end up with is a bunch of songs from a bunch of different discs at a bunch of different volume levels. I read that there is a feature to normalize the volume on music playback but I don't see this feature for video playback. One thought was to strip the audio from the hdmi and run it through a compressor then back into my AVR as an analog signal. Does anyone have any other tricks up their sleeve to normalize the volume on video playback when the songs come from many different concert videos?

Thanks, Dean

 

[Transepoch]

You may want to see if that will work anyway. The setting mentions music, but having it set greys the option under the player's Audio tab.

I would think player decoding and then imposing dynamic range controls could also keep levels leashed. Some receivers also have a similar function (even without forcing Dolby Night mode on.)

 

[ddetton]

Good idea! Kscape support did not suggest that but Andre did suggest player decoding with DRC. I tried the player decoding with DRC and there is still a pretty wide range of volume levels. I'll try the setting related to music and see if that helps. Thanks!

 

[sbutter]

At Celia, I noticed some demo scripts used the Script step ‘send command to control system’ to set volume level before each scene played.

 

[ddetton]

That’s an interesting idea. Thanks for your input. The tricky part is that it would have to be a value in reference to everything else. Just as an example I could use a rating system from 1 to 10 with 1 being very quiet and 10 being extremely loud. And since it is random playback you have to look at the rating of the last song that played to determine if you need to raise or lower the volume. I think that you’d have to have a target rating like 5. If it’s higher, then lower the volume and if it’s lower then raise the volume. It’s a tough problem to solve using programming.
My next step is to run the analog out of the M500 into a pro level compressor and then the output of the compressor back into an analog in on the receiver. I am hoping that I don’t have to jump up to the really expensive ones to get it to sound right. Yes, you do lose any multi-channel but in this application multi-channel is not that important.

 

[MikeKobb]

I can save you some time: the volume adjustment feature only works on playback of imported CDs.

The issue with video content is that there are multiple audio codecs, many of which are output as a bitstream. That makes it impossible for us to adjust the volume without internally decoding the audio, and then potentially re-encoding it for output, which is not something we're able to do.

Also, volume normalization requires that we have information about the volume levels present in the content. This is determined for CDs during the import process, but nothing like that is done for video content. If we ever wanted to support volume normalization on video content, we'd have to go back and retroactively decode all of the audio on all of the content on the system to measure its levels.

 

[ddetton]

I’ve been wondering how that works. Thanks for the behind the scenes explanation @MikeKobb ! Sounds like the team responsible for decoding the video content never anticipated that anyone would try to do what I’m doing . One more thing for me to try … connect a compressor to the analog out of the m500 to see if I can solve this issue without totaling destroying the audio. Thanks Mike!

 

[o0OBillO0o]

I can save you some time: the volume adjustment feature only works on playback of imported CDs.

The issue with video content is that there are multiple audio codecs, many of which are output as a bitstream. That makes it impossible for us to adjust the volume without internally decoding the audio, and then potentially re-encoding it for output, which is not something we're able to do.

Also, volume normalization requires that we have information about the volume levels present in the content. This is determined for CDs during the import process, but nothing like that is done for video content. If we ever wanted to support volume normalization on video content, we'd have to go back and retroactively decode all of the audio on all of the content on the system to measure its levels.

Wow! I see that supporting live decoding and offset measurement could be a non starter from a cost/licensing perspective. Let alone processing power.

When Kaleidescape processes the mezzanine file, what type of effort would it take to capture the offset value from, say, a predetermined level and embed as meta data?

Said differently, 0 dBFS could be the reference level , a movie file would have an OFFSET_VIDEO_AUDIO meta data value and a premium partner such as Crestron or Savant or Control4 could access this value to send the command to adjust/trim the volume level of the preamp/processor. Make sense?

Alternatively, we as users could attempt to measure the offset value for each disc or film and hardcode that command in Crestron SIMPL (whoa time suck!). I wonder if there's Disc database out there that captures this?

Interesting post and question!

 

[ddetton]

Wow! I see that supporting live decoding and offset measurement could be a non starter from a cost/licensing perspective. Let alone processing power.

When Kaleidescape processes the mezzanine file, what type of effort would it take to capture the offset value from, say, a predetermined level and embed as meta data?

Said differently, 0 dBFS could be the reference level , a movie file would have an OFFSET_VIDEO_AUDIO meta data value and a premium partner such as Crestron or Savant or Control4 could access this value to send the command to adjust/trim the volume level of the preamp/processor. Make sense?

Alternatively, we as users could attempt to measure the offset value for each disc or film and hardcode that command in Crestron SIMPL (whoa time suck!). I wonder if there's Disc database out there that captures this?

Interesting post and question!

Normalizing volume levels for video playback would be a whole lot easier for Kscape to do than for the community. I understand why it might not get done for the Premiere line but is something that should definitely be considered for the current product line. I like your idea of establishing a reference level for the volume then storing the offset as meta data For each disc. The user can then choose to use it or not. If there is no such database out there, it’s not a bad idea to start one as an open source project. I’m going to look around first. I’m also going to look at just the content downloaded from the kstore to see if that volume level is normalized.

 

[o0OBillO0o]

Normalizing volume levels for video playback would be a whole lot easier for Kscape to do than for the community. I understand why it might not get done for the Premiere line but is something that should definitely be considered for the current product line. I like your idea of establishing a reference level for the volume then storing the offset as meta data For each disc. The user can then choose to use it or not. If there is no such database out there, it’s not a bad idea to start one as an open source project. I’m going to look around first. I’m also going to look at just the content downloaded from the kstore to see if that volume level is normalized.

Here's an old list. I didn't find the methodology..

more here..

Understanding Dialog Normalization - Blu-ray Forum

Understanding Dialog Normalization Audio Theory and Discussion

forum.blu-ray.com