Encoding Lossless Audio to E-AC3 5.1

[FoodOcean]

So, I am entertaining the idea of making 5.1 E-AC3 versions of all of my lossless tracks, for compatibility and space saving purposes. I would be to keep any 7.1 tracks, as handbrake can only passthrough those, any 5.1 tracks would be replaced by an E-AC3 encode of the highest quality/bitrate audio track available.

With that said, I have a few of questions that I wanted to run by y'all before I dive headfirst into updating the audio for all my content:

1. I read that that the audio encoder for FFMPEG/Handbrake is not the best when it comes to Dolby codecs because of their licensing. Id have to jump through some hoops, but I may have the opportunity to get a trial license for Dolby Media Encoder; however, I would need to remix the dts-ma track to a wav file, explode it into mono channels, and then re-encode then back together into a surround e-ac3 track. is it worth it and if so, does anyone know the optimal settings to do so?

2. Am I correct in thinking that 5.1 E-AC3 @ 640kbps sounds transparent to Lossless 5.1 in most situation? For context, I have a Samsung HW-Q990D soundbar system. It's good, but it's not studio/home theater levels of fidelity.

3. Is it better to make an 5.1 E-AC3 version out of the lossless track, or just use its core (i.e. - DTS @ 1536kbps or AC3 @ 640kbps)?

4. This question, I feel like I already know, but I just want to confirm: If there are a 7.1 and 5.1 lossless track, I should encode the 5.1 to E-AC3 to prevent any funny business occurring with the downmix, right?

[FranceBB]

1. I read that that the audio encoder for FFMPEG/Handbrake is not the best


You'd be surprised by how good the open source encoders have become, especially the normal AC3 one and they can even insert the proprietary metadata like the dialnorm etc.


2. Am I correct in thinking that 5.1 E-AC3 @ 640kbps sounds transparent to Lossless 5.1 in most situation?


That entirely depends on your ears as everyone is different, but in my case if I'm using my 5.1 I wouldn't notice the difference. You can still make a test at various bitrates, check with Spek (drag and drop the two audio tracks, the orignal and the re-encoded one) and see the difference in the spectrum and how the frequencies are affected.

By the way, regular broadcast tv uses 384 kbit/s 48000Hz hardware encoded AC3 for 5.1 and most people are fine watching movies and tv series this way, so a 640 kbit/s E-AC3 should be plenty. :)


3. Is it better to make an 5.1 E-AC3 version out of the lossless track, or just use its core?


Good question. I would personally re-encode if I had to as it would be my personal preference, but I'll leave it to other people like Tebasuna to comment on this one.


4. This question, I feel like I already know, but I just want to confirm: If there are a 7.1 and 5.1 lossless track, I should encode the 5.1 to E-AC3 to prevent any funny business occurring with the downmix, right?

I only have a 5.1 and I always take the 5.1 to produce a 5.1 output so that I know that no further processing occurs.

[tebasuna51]

A curious soundbar with a separate subwoofer and surround sound system featuring 23 speakers.
For under €800, we can't expect high fidelity, but we can expect an immersive 3D experience.

I would recommend preserving Atmos tracks when they contain real 3D information.

However, 7.1 2D tracks can be replaced with 5.1 tracks without losing any information.

We need to know the player. If it's the TV directly, the only compatible format will probably be Dolby. The soundbar doesn't support DTS, and although it claims to support FLAC and AAC, I'm not sure if they can be transmitted via HDMI with your player (with my PC I send PCM multichannel decoding any source, then I use AAC with better quality/bitrate).

1 and 2) Like your audio system is not a HiFi system we can't expect differences between the encoder (Dolby or ffmpeg) and a 640 Kb/s can be transparent for you. BTW you can test yourself between the encoders and bitrate like FranceBB say. If you need monowavs to test with Dolby Encoder remember eac3to:

eac3to INPUT_TRACK output.wavs

3 and 4) If you have a lossless DTS 7.1 you can use the core 1536 Kb/s 5.1 to obtain the 640 Kb/s EAC3 instead downmix yourself the 7.1.
If you have a THD 7.1 from a BD you can use the companion AC3 directly, from a THD 5.1 you can recode it to EAC3.
Basically, I prefer an original 5.1 audio to a manual downmix.

[FoodOcean]

I would recommend preserving Atmos tracks when they contain real 3D information.

However, 7.1 2D tracks can be replaced with 5.1 tracks without losing any information.

I have no Atmos tracks yet, just TrueHD and DTS-HD MA (in 5.1 and 7.1)

We need to know the player. If it's the TV directly, the only compatible format will probably be Dolby. The soundbar doesn't support DTS, and although it claims to support FLAC and AAC, I'm not sure if they can be transmitted via HDMI with your player (with my PC I send PCM multichannel decoding any source, then I use AAC with better quality/bitrate).

The sound bar does support DTS, but my tv does not. I plan to get a different tv or a tv stick that will support dts, but that's a way out.

1 and 2) Like your audio system is not a HiFi system we can't expect differences between the encoder (Dolby or ffmpeg) and a 640 Kb/s can be transparent for you.

I can hear the difference in headphones, but I have dried the sound system yet.

If you need monowavs to test with Dolby Encoder remember eac3to:
eac3to INPUT_TRACK output.wavs

can't FFMPEG do that as well? is eac3to better?

3 and 4) If you have a lossless DTS 7.1 you can use the core 1536 Kb/s 5.1 to obtain the 640 Kb/s EAC3 instead downmix yourself the 7.1.
If you have a THD 7.1 from a BD you can use the companion AC3 directly, from a THD 5.1 you can recode it to EAC3.
Basically, I prefer an original 5.1 audio to a manual downmix.

isnt it better to encode from a lossless source though? do you think the included ac3 would be better than a DME eac3?

[FoodOcean]

That entirely depends on your ears as everyone is different, but in my case if I'm using my 5.1 I wouldn't notice the difference. You can still make a test at various bitrates, check with Spek (drag and drop the two audio tracks, the original and the re-encoded one) and see the difference in the spectrum and how the frequencies are affected.

where can I do that? what tool do I use? Also, I agree the encoder isn't that bad, like the AAC one is significantly better than people give it credit for. I do know that the DME encoder for Dolby stuff is better, I can hear the difference

[FranceBB]

where can I do that?

I meant something like this for AC3:


ffmpeg.exe -i "U2673397.avs" -c:a ac3 -b:a 384k -ar 48000 -center_mixlev 0.500 -surround_mixlev 0.500 -dialnorm -23 -dmix_mode loro -ltrt_cmixlev 0.707 -ltrt_surmixlev 0.707 -loro_cmixlev 0.707 -loro_surmixlev 0.707 -y "test.ac3"

pause


and like this for E-AC3:


ffmpeg.exe -hide_banner -i "U3031512.avs" -c:a eac3 -b:a 384k -ar 48000 -dialnorm -23 -dmix_mode loro -ltrt_cmixlev 0.707 -ltrt_surmixlev 0.707 -loro_cmixlev 0.707 -loro_surmixlev 0.707 -y "test.eac3"

pause


By the way, for 5.1 at 48000Hz E-AC3 can go all the way up to 6.1 Mbit/s, in other words -b:a 6144k. My suggestion is to try with 384 kbit/s and then go up until you hit the sweet spot, which in your case should be 640 kbit/s as you were saying.

The open source encoder works well these days.
If you need updated builds of FFMpeg and you don't wanna build them yourself, you can find them here: https://github.com/BtbN/FFmpeg-Builds/releases

Pick the ones under GPL license and you're good to go. ;)

As to Spek, this is available here https://github.com/alexkay/spek/releases which is also open source and it's very easy to use, just open it and drag and drop the original audio track and the encoded audio track to see the differences in the spectrum.

https://images2.imgbox.com/d2/fa/KvlvUtw4_o.png




I agree the encoder isn't that bad, like the AAC one is significantly better than people give it credit for.


Well, for AAC it depends on *which* encoder. FDK_AAC from Fraunhofer is generally the way to go. The one included in libavcodec is actually pretty poor to this day. It got a bit better, but it's still fairly poor and requires incredibly high bitrates.

[FoodOcean]

Well, for AAC it depends on *which* encoder. FDK_AAC from Fraunhofer is generally the way to go. The one included in libavcodec is actually pretty poor to this day. It got a bit better, but it's still fairly poor and requires incredibly high bitrates.

is fdk better than qaac? also is there a way to incorporate either of them into a handbrake/ffmpeg GUI?

Secondly, do you know the mixdown equation/balancing the that handbrake defaults to? I have been trying to recreate it in FFMpeg to no luck!

[kurkosdr]

You'd be surprised by how good the open source encoders have become, especially the normal AC3 one and they can even insert the proprietary metadata like the dialnorm etc.

Which do you consider the best open-source AC3 encoder?

FDK_AAC from Fraunhofer is generally the way to go. The one included in libavcodec is actually pretty poor to this day. It got a bit better, but it's still fairly poor and requires incredibly high bitrates.
The worst kind of not-invented-here is the open-source kind, I guess. :(

(note: This doesn't apply to AC3 since Dolby never provided a reference AC3 encoder... a compression standard only has to specify the decoder and providing a reference encoder is a courtesy, but AAC does have a very good reference encoder in the form of FDK_AAC, shame it's not used by libavcodec )

[tebasuna51]

The sound bar does support DTS, but my tv does not. I plan to get a different tv or a tv stick that will support dts, but that's a way out.
Then we need EAC3 always.
I use AAC with QAAC encoder to 5.1 and ffmpeg-fdkaac to 3D 5.1.2

I can hear the difference in headphones, but I have dried the sound system yet.
Then use the Dolby Media Encoder if you have it.

can't FFMPEG do that as well? is eac3to better?
In order to decode DTS-MA or TrueHD tracks is the same decoders, only is a easy sintax.

isnt it better to encode from a lossless source though? do you think the included ac3 would be better than a DME eac3?
It is better use the lossless source of course, but in order to downmix 7.1 to 5.1 use something like (https://forum.doom9.org/showthread.php?p=2017559#post2017559):
ffmpeg -hide_banner -drc_scale 0 -bitexact -i INPUT_71 -vn -filter_complex^
"asplit [f][s]; [f] pan=3.1|c0=c0|c1=c1|c2=c2|c3=c3 [r]; [s] pan=stereo|c0=0.5*c4+0.5*c6|c1=0.5*c5+0.5*c7, compand=attacks=0:decays=0:points=-90/-84|-8/-2|-6/-1|-0/-0.1, aformat=channel_layouts=stereo [d]; [r][d] amerge [a]"^
-map "[a]" -acodec pcm_s24le -f wav OUTPUT_51.wav

[FoodOcean]

Then we need EAC3 always.
I use AAC with QAAC encoder to 5.1 and ffmpeg-fdkaac to 3D 5.1.2

I use AAC for stereo, but I have heard the default encoder in ffmpeg and handbrake aren't great, and that I should use QAAC or FDK. Do you know the default downmix prompt for handbrake? I like handbrake's default downmix algorithm (regardless of the default encoder), so I am trying to reverse engineer it in ffmpeg, but I can't find the right settings/prompt to use.

[FranceBB]

Which do you consider the best open-source AC3 encoder?


The one in FFMpeg which is part of libavcodec as it's the one that is kept most updated. After that, there's the old Aften encoder which I've used in the past to create DVD via AVS to DVD. Nowadays, the AC3 encoder in FFMpeg can be used to create a proper stream and even official decoders read the metadata correctly, accept it and apply the right DRC etc.

I don't have much time these days, but one of the things I have planned is making a cycle encode test using the Dolby DP600 from 2007 and the current open source AC3 encoder in FFMpeg. I don't think they're a million miles apart, if anything, I think they're pretty close right now.


This doesn't apply to AC3 since Dolby never provided a reference AC3 encoder... a compression standard only has to specify the decoder and providing a reference encoder is a courtesy, but AAC does have a very good reference encoder in the form of FDK_AAC, shame it's not used by libavcodec

to give you an idea of how bad the AAC encoder in libavcodec is, last time I made a comparison in FFMpeg 6.1.1, a 384 kbit/s AAC 5.1 48000Hz track sounded much worse than a 384 kbit/s AC3 5.1 48000Hz track from the Dolby DP600, while actually it should be the other way round as AAC is supposed to be slightly more efficient.

is fdk better than qaac?

In order of quality:

1) Apple AAC AudioToolbox (qaac)
2) Frauhnofer AAC (fdk aac)
3) Nero AAC
4) FFMpeg AAC (lavf / libavcodec aac)

[FoodOcean]

In order of quality:

1) Apple AAC AudioToolbox (qaac)
2) Frauhnofer AAC (fdk aac)
3) Nero AAC
4) FFMpeg AAC (lavf / libavcodec aac)

Ok, and do you know how handbrake downmixes? I know Dolby standards is to reduce all the channels by -3db and fold them into L and R respectively, is that he case as well?

Also, I have tried the QAAC encoder and when I used average bitrate at 160kbps for stereo, it gives me 145kbps some reason. Versus libavcodec will hit that target. is it because QAAC is nor efficient? and can do the same/better quality at lower bitrate?

[tebasuna51]

Ok, and do you know how handbrake downmixes?
I try Handbrake 1.8.2 (last version not supported in my W10).
The output for a 7.1 -> 5.1 is:
FL' = FL
FR' = FR
FC' = FC
LFE'= LFE
SL' = 0.707*BL + SL
SR' = 0.707*BR + SR
Like you can see the Surround channels can be clipped.
I recommend the method in my previous post with:
SL' < BL + SL
SR' < BR + SR
With < means Normalized to max value 1.0 instead 1.707 only for high values.

Also, I have tried the QAAC encoder and when I used average bitrate at 160kbps for stereo, it gives me 145kbps some reason.

You must use CVBR with qaac encoder for average variable bitrate, if the output have less bitrate maybe reach the limit and don't need more.

In order to encode to EAC3 I make a GUI for Dolby Encoder Engine (https://forum.doom9.org/showthread.php?p=1995743#post1995743), maybe I can add the downmix method when convert any source 7.1 to RF64 format needed by DEE.

[FoodOcean]

I try Handbrake 1.8.2 (last version not supported in my W10).
The output for a 7.1 -> 5.1 is:
FL' = FL
FR' = FR
FC' = FC
LFE'= LFE
SL' = 0.707*BL + SL
SR' = 0.707*BR + SR
Like you can see the Surround channels can be clipped.
I recommend the method in my previous post with:
SL' < BL + SL
SR' < BR + SR
With < means Normalized to max value 1.0 instead 1.707 only for high values.

I am looking to downmix to stereo from 5.1 for AAC, so you are saying that I should do FL' < SL + FL + FC and FR' < SR + FR + FC? does Handbrake do L' = 0.707(SL + FC) + FL and R' = 0.707(SR + FC) + FR?



You must use CVBR with qaac encoder for average variable bitrate, if the output have less bitrate maybe reach the limit and don't need more.

I will try it out!

In order to encode to EAC3 I make a GUI for Dolby Encoder Engine (https://forum.doom9.org/showthread.php?p=1995743#post1995743), maybe I can add the downmix method when convert any source 7.1 to RF64 format needed by DEE.

that is so cool! I wish the dialnorm had an Auto option though

[tebasuna51]

New DeeGUI:
Added function to downmix the obsolete (I think) 2D 7.1 surround layout to the equivalent 5.1

Use 0 to Auto DialNorm.

[FoodOcean]

New DeeGUI:
Added function to downmix the obsolete (I think) 2D 7.1 surround layout to the equivalent 5.1

Use 0 to Auto DialNorm.

Thanks!
is there a reason you recommend keeping the drc on "music light" and not "film Light"?

[j7n]

that is so cool! I wish the dialnorm had an Auto option though
When encoding from the command-line from XML, there is an auto option. I believe it is a default. It performs 2 passes, first measure then encode. It basically sets dialnorm to the LUFS loudness when encoding music, and attempts to extract dialog in films (not sure how accurate that is on a full mix). I feel that this is rather wrong because isolated music should be louder than dialog.


<loudness> <!-- measure_only must be used in DD mode. -->
<measure_only>
<metering_mode>1770-3</metering_mode> <!-- One of: 1770-3, 1770-2, 1770-1, LeqA -->
<dialogue_intelligence>true</dialogue_intelligence> <!-- boolean: true or false -->
<speech_threshold>20</speech_threshold> <!-- integer: from 0 to 100 -->
</measure_only>
</loudness>

[FoodOcean]

When encoding from the command-line from XML, there is an auto option. I believe it is a default. It performs 2 passes, first measure then encode. It basically sets dialnorm to the LUFS loudness when encoding music, and attempts to extract dialog in films (not sure how accurate that is on a full mix). I feel that this is rather wrong because isolated music should be louder than dialog.


<loudness> <!-- measure_only must be used in DD mode. -->
<measure_only>
<metering_mode>1770-3</metering_mode> <!-- One of: 1770-3, 1770-2, 1770-1, LeqA -->
<dialogue_intelligence>true</dialogue_intelligence> <!-- boolean: true or false -->
<speech_threshold>20</speech_threshold> <!-- integer: from 0 to 100 -->
</measure_only>
</loudness>


Am I reading that code right in that the "measure" function should only be used for DD content?

[j7n]

This is an excerpt of the preset. There is also a <measure_and_correct> mode, which will normalize to a target with limiting instead of just measuring. It says that correction cannot be done in DD mode, only in DDP (E-AC-3).

[tebasuna51]

is there a reason you recommend keeping the drc on "music light" and not "film Light"?
Because I don't know how to force DRC to 'none', and light music is the closest.

I don't like Dolby's volume control system based on DN and DRC.
If all audio were Dolby, it would be useful, but there's also CD audio (loudness war (https://en.wikipedia.org/wiki/Loudness_war)), TV commercials, and other codecs that produce much higher volumes and make that system unfavorable.

Therefore, I use ffmpeg as my encoder (DN=-31, DRC=none) or AAC.

[FoodOcean]

Because I don't know how to force DRC to 'none', and light music is the closest.

I don't like Dolby's volume control system based on DN and DRC.
If all audio were Dolby, it would be useful, but there's also CD audio (loudness war (https://en.wikipedia.org/wiki/Loudness_war)), TV commercials, and other codecs that produce much higher volumes and make that system unfavorable.

Therefore, I use ffmpeg as my encoder (DN=-31, DRC=none) or AAC.

Oh that’s so interesting, I didn’t even know this graph existed! Is that a graph you found, or one that you made based on your analysis/research?

You say you use ffmpeg as your primary encoder, do you find the default/open source encoder good enough for your needs? I only started looking into DME because most information I’ve received says that the default encoder in ffmpeg and handbrake are mediocre when it comes to AAC and Dolby codecs

[FoodOcean]

This is an excerpt of the preset. There is also a <measure_and_correct> mode, which will normalize to a target with limiting instead of just measuring. It says that correction cannot be done in DD mode, only in DDP (E-AC-3).

Do you access the CLI via CMD? I’ll have to check it out! Rn I’m tryna figure out the most optimal way to encode dts-hd and THD to E-AC3 without changing their baked in mastering and loudness balance. I guess that’s always the goal with encoding though huh, as close to the original while reducing size lol

[FranceBB]

Because I don't know how to force DRC to 'none', and light music is the closest.


You mean in the open source encoder? Interesting.
I can definitely see it in the Dolby proprietary one and I can change it from Film Light to None

https://images2.imgbox.com/53/fa/5KuDZS11_o.pnghttps://images2.imgbox.com/6a/e4/jUMaMlAU_o.png

and sure enough on the decoder side it can see when it's not set, in fact here is how the decoder reads it when it's Film Light and how it reads it when it's set to none

https://images2.imgbox.com/a1/52/f7OV855V_o.pnghttps://images2.imgbox.com/80/b2/udk8kCH0_o.png

which means that it's definitely an officially supported value for both AC3 and DolbyE. We just need a way for the open source encoder to signal it as None.

[j7n]

Do you access the CLI via CMD?
I access DEE from within Frontah, which is a customizable front-end to which many encoders can be added.

dee.exe --xml d:\apps\dee\xml\ddp.xml -a '%i' -o '%o' --add-elem data_rate=%x %k

Here the variables are input, output, bitrate and custom parameters. XML is a prepared preset with typical settings. add-elem can override any of them.

Indeed, there isn't an option to set dynamic compression to none.

[tebasuna51]

I don't remember where I found that chart, but it can be done using the official data at https://professionalsupport.dolby.com/s/article/A-Guide-to-Dolby-Metadata?language=en_US

The open source encoder (ffmpeg) never adds DRC metadata; it is always equivalent to the 'none' preset on Dolby encoders.
I always add the parameter -center_mixlev 0.707, equivalent to Dolby default -3 dB, because ffmpeg default to 0.5 with low center volume when downmix to stereo.

As j7n has confirmed, with the 'dee' (Dolby Encoder Engine) it is not possible, or we don't know how, to choose the 'none' preset.
Although it is officially a valid option, perhaps with the encoder used by FranceBB.

My GUI create a 0.xml file, with the settings selected, and send a command equivalent to the j7n one.

[j7n]

The older encoding libraries available in software such as Sonic Foundry Vegas, Adobe Audition, MainConcept Total Code did have a "none" option. I think it should result in ever so slightly higher quality because the compression metadata doesn't occupy space in the bitstream.

I think center reduced by 3 dB has always been a good default. The center now plays through two stereo speakers, which sum imperfectly in a space.

[FoodOcean]

https://images2.imgbox.com/53/fa/5KuDZS11_o.pnghttps://images2.imgbox.com/6a/e4/jUMaMlAU_o.png

https://images2.imgbox.com/a1/52/f7OV855V_o.pnghttps://images2.imgbox.com/80/b2/udk8kCH0_o.png


What version of the DME/DEE are you using? My GUI doesn't look like that, it only shows: Film light, Film Standard, Music Light, Music standard, and speech.

[FoodOcean]

I access DEE from within Frontah, which is a customizable front-end to which many encoders can be added.

dee.exe --xml d:\apps\dee\xml\ddp.xml -a '%i' -o '%o' --add-elem data_rate=%x %k

Here the variables are input, output, bitrate and custom parameters. XML is a prepared preset with typical settings. add-elem can override any of them.

Indeed, there isn't an option to set dynamic compression to none.

ok, I will check this out! do you feel it gives you more/better control than the DME GUI? like is it worth it?

[FoodOcean]

The open source encoder (ffmpeg) never adds DRC metadata; it is always equivalent to the 'none' preset on Dolby encoders.
I always add the parameter -center_mixlev 0.707, equivalent to Dolby default -3 dB, because ffmpeg default to 0.5 with low center volume when downmix to stereo.

I think handbrake's default downmix is also 0.707 vs ffmpeg's default 0.5, I can't tell for sure, but when I listen to them then sound the same vs the default ffmpeg one.

[FoodOcean]

The older encoding libraries available in software such as Sonic Foundry Vegas, Adobe Audition, MainConcept Total Code did have a "none" option. I think it should result in ever so slightly higher quality because the compression metadata doesn't occupy space in the bitstream.

are you using the olde libraries, which is why you can access the "none"?

[j7n]

do you feel it gives you more/better control than the DME GUI?
I use old tools generally. I am not experienced with DEE GUI. It often goes into Script Paused mode and stops working. Looks like it already provides an "Auto" dialnorm if you specify 0, but I am not able to test it. When calling DEE directly, I need to supply it an RF64 or a WAV file, which DEE GUI would produce for you.

The way I DEE, dee.exe must be in the system path, and an absolute path to an XML needs to be provided. I can send you my customized Frontah if you want. The altered part is just a zip file with text.

I confirm with MediaInfo (full parsing, advanced) that there are no compr variables in the stream from TotalCode.

Here is a screenshot from TotalCode and spectra produced by all encoders. Nothing clearly stands out apart from the pro encoders using channel coupling in the high range at 384 kbit/s, and their strategies are slightly different. Ffmpeg doesn't seem to use coupling, but the impact of it can't be seen from the spectrogram.

https://imgur.com/a/Vq5vl7u

[tebasuna51]

Yes, I test TotalCode (https://forum.doom9.org/showthread.php?p=1967227#post1967227) and Audition 2017 (https://forum.doom9.org/showthread.php?p=1963144#post1963144) both with the 'None' option

The old free Aften encoder doesn't have channel coupling, but I think the encoder included in ffmpeg (https://forum.doom9.org/showthread.php?p=1794293#post1794293) (from the same author) does.

[tebasuna51]

I am not experienced with DEE GUI. It often goes into Script Paused mode and stops working.

My GUI need to be located at same folder than dee.exe with also with ffmpeg (or ffmpeg in SYSTEM path) and:

1) Create a INPUT_RF64.wav at same folder than INPUT file.
Is that file created on your system?

2) Create a 0.xml with the settings in dee.exe folder.
Is that file created on your system?

3) Now run:
dee.exe -x 0.xml --verbose info --log-file 0.log --progress
Seems work dee.exe and create the 0.log?

EDIT: maybe if you have dee.exe in a system folder my GUI can't create files (0.xml, o.log) in that folder.

Looks like it already provides an "Auto" dialnorm if you specify 0, but I am not able to test it.
Yes that work.

[j7n]

ffmpeg didn't seem to use coupling in my test because there was no bleed from the front into the rear channels at all at 384 kbit. But usually we would not select a low bitrate where that is needed. In AC-3 at 640 no coupling is used, as reported by AC3Filter (20/20).

DEE GUI locks up before I can start encoding, sometimes I can use the Open button once. Then neither Help nor the system close button work, and it must be closed from the taskbar. I tried to place it in the same folder as DEE. But it is not worth debugging for me because I don't need to use DEE GUI. It is possibly caused by my system being Server 2008 R2.

[FoodOcean]

I use old tools generally. I am not experienced with DEE GUI. It often goes into Script Paused mode and stops working. Looks like it already provides an "Auto" dialnorm if you specify 0, but I am not able to test it. When calling DEE directly, I need to supply it an RF64 or a WAV file, which DEE GUI would produce for you.
https://imgur.com/a/Vq5vl7u

I guess I am still a bit confused as to how you are able to use "None" as you profile. Would you be open to me reaching out to you via DM?

[j7n]

I can't choose "None" in Dolby Encoding Engine. The screenshot was from MainConcept TotalCode.

[FoodOcean]

I can't choose "None" in Dolby Encoding Engine. The screenshot was from MainConcept TotalCode.

and that program is an official authorized Dolby encoder?

[j7n]

I am almost certain that it has an encoder library licensed from Dolby as does other paid software.

[FranceBB]

What version of the DME/DEE are you using? My GUI doesn't look like that, it only shows: Film light, Film Standard, Music Light, Music standard, and speech.

That's not DEE, it's Emotion Engine which uses the proprietary Dolby SDK under the hood.
It's what we're using alongside the Dolby DP600, the glorious machines you can see here and that I saved two summers ago when the disks broke:

https://images2.imgbox.com/cd/ab/e6TjAS7B_o.png


Considering those things are from 2007, I wouldn't be surprised if it got removed in newer versions.

[FoodOcean]

I got into contact with Dolby tech support, and they said, " while the 'none' profile is technically in the source code of the encoder, it is not reachable in [their] commercial products" (Which explains why other authorized Dolby encoders may provide access to it in their suites).

the rep did say though that if you disable (or just don't enable) DRC in your client device/receiver, the then the metadata won't be read, and the sound mixing should be no different than the source.

[tebasuna51]

...the rep did say though that if you disable (or just don't enable) DRC in your client device/receiver, the then the metadata won't be read, and the sound mixing should be no different than the source.
The problem is that decoders/players apply DRC by default, forcing you to explicitly disable it.

In both ffmpeg and AviSynth decoders, you have to disable it if you want to recover the original sound.

[j7n]

I stumbled upon discussions where it was requested that drc_scale should be off, and the responses were basically telling "you don't know what you are doing" and "prove that it sounds better" and a lot of time wasting on arguments ....

https://forum.audacityteam.org/t/request-ffmpeg-plugin-with-drc-scale-disabled/58199

https://ffmpeg.org/pipermail/ffmpeg-devel/2015-March/170758.html

I have to add drc_scale 0 to every ffmpeg preset I use because Dolby spec says that DRC should be on by default.

The older Dolby encoders mentioned earlier are also "commercial prodcuts". Only in the latest DEE the option is inaccessible.

[FoodOcean]

I know 640kbps is considered “transparent” for 5.1, but what bitrate is considered transparent for 7.1? I’ve read it can be as low as 768kbps and has high as 1mbps.

[microchip8]

I know 640kbps is considered “transparent” for 5.1, but what bitrate is considered transparent for 7.1? I’ve read it can be as low as 768kbps and has high as 1mbps.

I usually use 960 kbps for 7.1

[FoodOcean]

I usually use 960 kbps for 7.1

Is that with or without atmos?

[microchip8]

Is that with or without atmos?

Without.

I don't encoder or copy Atmos because I find its bitrate too high, sometimes shooting up to 4 Mbps which hovers close to that of the video itself (for the most of my encodes). So, just for audio 4 or 4.5 Mbps is just too high for me so I encode everything to AC-3 @ 640 kbps or copy the core of DTS in case the track is DTS-HD MA (core is most of the time 1.5 Mbps)

[FoodOcean]

I just ran some tests and while the “streaming” profile of E-AC3 goes up to 1024kbps, it is limited to 5.1. If you want to do 7.1, you have to do the “Blu-ray” profile, which goes up to 1.6mbps (at least using DME).

[Balling]

I stumbled upon discussions where it was requested that drc_scale should be off, and the responses were basically telling "you don't know what you are doing" and "prove that it sounds better" and a lot of time wasting on arguments ....

https://forum.audacityteam.org/t/request-ffmpeg-plugin-with-drc-scale-disabled/58199

https://ffmpeg.org/pipermail/ffmpeg-devel/2015-March/170758.html

I have to add drc_scale 0 to every ffmpeg preset I use because Dolby spec says that DRC should be on by default.

The older Dolby encoders mentioned earlier are also "commercial prodcuts". Only in the latest DEE the option is inaccessible.

DRC certainly should not be on by default when reencoding e.g. to flac or TrueHD even; it should also be disabled when reencoding to eac3 again. What Dolby is talking about is really the misunderstaning in the specification. See bigger discussion here, basically specification said this would be less CPU intensive. But we are past that for a long time, modern CPUs are very wide and very low latency... https://patchwork.ffmpeg.org/project/ffmpeg/patch/20200201193443.22419-1-rcombs@rcombs.me/

Mpv that uses ffmpeg disables drc. Lavfilters implemented option to apply DRC scale 0 in runtime. Dolby Reference decoder has option to disable DRC on eac3 or on TrueHD, yes it supports TrueHD magic stuff too... All normal AVRs support disabling DRC, night mode is useful, I guess.

[tebasuna51]

https://patchwork.ffmpeg.org/project/ffmpeg/patch/20200201193443.22419-1-rcombs@rcombs.me/

-1 for Derek Buitenhuis
+1 for Paul B Mahol

[hellgauss]

I hate metadata.

Metadata = I steal your remote-control and I tune your volume (and brightness) as I wish.