Atmos finally decoded in both PC/Mac - Page 3

Balling · 19th March 2023, 17:13

Quote:

Originally Posted by tebasuna51

That is very interesting. How can cancel DRC?

cd to DRP folder and run using ./ (absolute path will not work):

./gst-inspect-1.0.exe --gst-plugin-path gst-plugins dlbac3dec

Compare to EAC3 interface in DRP, the app: drc-cut, drc-boost and drc-mode are the same as there. Set to true drc-suppress.

It also has drop-delay: Drop delay samples added by the decoder at the stream start

Nice.

in ./gst-inspect-1.0.exe --gst-plugin-path gst-plugins dlbtruehddec

it is already disabled, but you can enable it here.

Obviously you can just do it though global dlbaudiodecbin, ac3dec-drop-delay, ac3dec-drc-suppress

I checked, options work

Balling · 19th March 2023, 20:34

Full command "C:\Program Files\Dolby\Dolby Reference Player\gst-launch-1.0.exe" --gst-plugin-path "C:\Program Files\Dolby\Dolby Reference Player/gst-plugins" filesrc location=C:\\tmp\\atmos\\gapless.eac3 ! dlbac3parse ! dlbaudiodecbin ac3dec-drop-delay=true ac3dec-drc-suppress=true out-ch-config=13 ! audio/x-raw,format=S32LE ! wavenc ! filesink location=Z:\zE.L111.wav

tebasuna51 · 22nd March 2023, 23:46

Ok, and for thd files with drc-suppress by default:

"C:\Program Files\Dolby\Dolby Reference Player\gst-launch-1.0.exe" --gst-plugin-path "C:\Program Files\Dolby\Dolby Reference Player/gst-plugins" filesrc location=C:\\tmp\\atmos\\zT.thd ! dlbtruehdparse align-major-sync=false ! dlbaudiodecbin truehddec-presentation=16 out-ch-config=13 ! audio/x-raw,format=S32LE ! wavenc ! filesink location=C:\\tmp\\atmos\\zT.wav

Y try format=S24LE and report a error like not supported, no problem with the output 32 int.
It is much more fast decode the 8 channels at same time, the problem is the wav filesize >4 GB (only can fit a time of 46:36:12).
Audition (and other soft) can't load that files with standard (and wrong) header, need a w64 or rf64 header. We can use:

eac3to zT.wav zT.w64

With automatic downsample to 24 int, enough for a eac3 encode.

Attached .bat and log

Balling · 23rd March 2023, 17:15

drc-suppress should be only for EAC3/AC3/AC4, THD has drc off by default. Yes, only int 24 inside int 32 is supported.

Eviluess · 17th August 2023, 07:30

Hi, have you noticed that the generated result has around 7dB lower volume than the audio converted by ffmpeg -i input.mp4 output.wav?

Quote:

Originally Posted by Balling

cd to DRP folder and run using ./ (absolute path will not work):

./gst-inspect-1.0.exe --gst-plugin-path gst-plugins dlbac3dec

Compare to EAC3 interface in DRP, the app: drc-cut, drc-boost and drc-mode are the same as there. Set to true drc-suppress.

It also has drop-delay: Drop delay samples added by the decoder at the stream start

Nice.

in ./gst-inspect-1.0.exe --gst-plugin-path gst-plugins dlbtruehddec

it is already disabled, but you can enable it here.

Obviously you can just do it though global dlbaudiodecbin, ac3dec-drop-delay, ac3dec-drc-suppress

I checked, options work

tebasuna51 · 17th August 2023, 09:17

Yep, 7dB or other, seems the drc-suppress is ignored by the decoder and the drc is applied always over eac3 Atmos.

I'm using now the eac3to 'normalize' after decode:

Quote:

set TEM=C:\\tmp\\atmos\\
set DECODER=C:\Program Files\Dolby\Dolby Reference Player
set EAC3TO=C:\Portable\eac3to\eac3to.exe
"%DECODER%\gst-launch-1.0.exe" --gst-plugin-path "%DECODER%/gst-plugins" filesrc location=%TEM%zE.eac3 ! dlbac3parse ! dlbaudiodecbin ac3dec-drop-delay=true ac3dec-drc-suppress=true out-ch-config=13 ! audio/x-raw,format=S32LE ! wavenc ! filesink location=%TEM%zE.wav
rem zE.wav to zE.w64 normalize and 32f to 24int (Audition can't read wav >4GB)
"%EAC3TO%" zE.wav zE.w64 -normalize

Eviluess · 19th August 2023, 11:08

Quote:

Originally Posted by tebasuna51

Yep, 7dB or other, seems the drc-suppress is ignored by the decoder and the drc is applied always over eac3 Atmos.

I'm using now the eac3to 'normalize' after decode:

Thank you for confirming this.
And as far as I know, any post-processing such as re-normalizing will compromise the "lossless" property. E-AC-3 is lossy, while TrueHD is lossless, so I hope TrueHD can do the job properly on respecting to the original.
Additionally, I found the TrueHD options allow us to use DRC modes, and only the value 4 can make the result louder than others. However, this make the end result LOUDER than the original, which introduces lossy in details.

Do you know how to make TrueHD without altering the original loudness?
Any suggestions would be greatly appreciated!
Thanks in advance!

tebasuna51 · 19th August 2023, 11:52

Quote:

Originally Posted by Eviluess

And as far as I know, any post-processing such as re-normalizing will compromise the "lossless" property. E-AC-3 is lossy...

You are right, the re-normalize is not the correct way, but is the best I found until we know a way to cancel the DRC applied by the decoder.

Quote:

... while TrueHD is lossless, so I hope TrueHD can do the job properly on respecting to the original.

The decoder default is not aply the DRC, then I also hope the decode is 'more or less' lossless.

BTW the atmos decode to a specific speaker configuration implies some lossy operations and we can't know how the original sound was recorded.

Eviluess · 19th August 2023, 14:48

Quote:

Originally Posted by tebasuna51

You are right, the re-normalize is not the correct way, but is the best I found until we know a way to cancel the DRC applied by the decoder.

The decoder default is not aply the DRC, then I also hope the decode is 'more or less' lossless.

BTW the atmos decode to a specific speaker configuration implies some lossy operations and we can't know how the original sound was recorded.

You make a good point. Let me provide some more details on my testing:
I used DaVinci Resolve to generate a test Atmos WAV file, playing simple sounds at each speaker channel like FL, FC, FR and so on. Then I used the Dolby Encoder Suite to generate TrueHD files from this.
When I compared the TrueHD output to the original WAV, I noticed the volume was lowered by about 7dB across all the speaker channels. This indicates a noticeable loss of fidelity to me, beyond what I'd consider "lossless." Boosting the volume with re-amping would also boost any lost details.
However, if having overhead channels is more important than absolute sound quality, this may be the "best" solution available right now, as you said.
In the end, it's a trade-off between full Atmos speaker support and lossless audio quality. I appreciate you helping me think through this - your insights are really valuable as I explore my options.
conclusions here!

Balling · 20th August 2023, 02:39

Quote:

Originally Posted by tebasuna51

You are right, the re-normalize is not the correct way, but is the best I found until we know a way to cancel the DRC applied by the decoder.

The decoder default is not aply the DRC, then I also hope the decode is 'more or less' lossless.

BTW the atmos decode to a specific speaker configuration implies some lossy operations and we can't know how the original sound was recorded.

TrueHD is not lossless, neither the Dolby Media encoder, nor the Dolby Reference player (hacked using the github script). FFmpeg decoder is lossless (and lossless with FFMpeg encoder), but FFmpeg encoder is slightly nonconformant at least with stereo.

Eviluess · 20th August 2023, 03:26

Quote:

Originally Posted by Balling

TrueHD is not lossless, neither the Dolby Media encoder, nor the Dolby Reference player (hacked using the github script). FFmpeg decoder is lossless (and lossless with FFMpeg encoder), but FFmpeg encoder is slightly nonconformant at least with stereo.

Since decoding the TrueHD to standard 7.1 surround PCM wave files doesn't introduce the perceived 7dB volume drop like the Dolby Reference Player does, I'm inclined to believe the Dolby Media Encoder didn't actually suppress the volume during encoding. Even if this process is lossy in some way, the end result seems acceptable to me.

It's still concerning that setting DRC to Off had no effect on fixing this issue. I would have expected that to bring the playback volume up to the proper level.

I tried using the ffmpeg command with the mlp/truehd encoder, and it lists max support up to 5.1 channels. This makes me think I may need to compile a custom version of ffmpeg with Dolby Atmos encoding enabled in order to get true lossless TrueHD output.

Code:

ffmpeg -h encoder=mlp

Eviluess · 23rd August 2023, 07:49

Quote:

Originally Posted by tebasuna51

You are right, the re-normalize is not the correct way, but is the best I found until we know a way to cancel the DRC applied by the decoder.

The decoder default is not aply the DRC, then I also hope the decode is 'more or less' lossless.

BTW the atmos decode to a specific speaker configuration implies some lossy operations and we can't know how the original sound was recorded.

BTW, do you know how to decode AC-4 to 16 channels with this tool?

tebasuna51 · 23rd August 2023, 13:18

Sorry, I don't know how.

Eviluess · 23rd August 2023, 15:21

Quote:

Originally Posted by tebasuna51

Sorry, I don't know how.

That's fine.
I don't know either.
And I tried combination of possible nodes but all failed.

Balling · 26th August 2023, 03:06

Quote:

Originally Posted by Eviluess

I tried using the ffmpeg command with the mlp/truehd encoder, and it lists max support up to 5.1 channels. This makes me think I may need to compile a custom version of ffmpeg with Dolby Atmos encoding enabled in order to get true lossless TrueHD output.

Code:

ffmpeg -h encoder=mlp

Why mlp?? ffmpeg -h encoder=truehd and ffmpeg -h decoder=truehd and decoder does support 7.1

"custom version of ffmpeg with Dolby Atmos encoding enabled" there is no Atmos support in THD or EAC3. No ac4 whatsover for now.

Eviluess · 3rd September 2023, 03:46

Quote:

Originally Posted by Balling

Why mlp?? ffmpeg -h encoder=truehd and ffmpeg -h decoder=truehd and decoder does support 7.1

"custom version of ffmpeg with Dolby Atmos encoding enabled" there is no Atmos support in THD or EAC3. No ac4 whatsover for now.

Thanks replying.
I just mentioned but I don't use ffmpeg to encode TrueHD or E-AC3. I use the Dolby Encoder Suite instead.

-QfG- · 13th October 2023, 17:03

How can i create a w64 file with GStreamer Plugin? If i use simply the *.w64 extension, the output file is corrupt:

Quote:

General
Format : Wave
Format settings : WaveFormatExtensible
File size : 8.93 GiB
Duration : 1 h 43 min
Overall bit rate mode : Constant
Overall bit rate : 12.3 Mb/s
FileExtension_Invalid : act at9 wav

Audio
Format : PCM
Format settings : Little / Signed
Codec ID : 00000001-0000-0010-8000-00AA00389B71
Duration : 1 h 43 min
Bit rate mode : Constant
Bit rate : 12.3 Mb/s
Channel(s) : 16 channels
Sampling rate : 48.0 kHz
Bit depth : 16 bits
Stream size : 8.93 GiB (100%)

if i convert this wav with FFMPEG to W64:

Quote:

General
Format : Wave64
Format settings : WaveFormatExtensible
File size : 948 MiB
Duration : 10 min 47 s
Overall bit rate mode : Constant
Overall bit rate : 12.3 Mb/s

Audio
Format : PCM
Format settings : Little / Signed
Codec ID : 00000001-0000-0010-8000-00AA00389B71
Duration : 10 min 47 s
Bit rate mode : Constant
Bit rate : 12.3 Mb/s
Channel(s) : 16 channels
Channel layout : L R C Lb Rb Cb Ls Rs Tfl Tfc Tfr Tbl Tbc Tbr
Sampling rate : 48.0 kHz
Bit depth : 16 bits
Stream size : 948 MiB (100%)

i have only 10 Minutes, lol. I can't play the WAV with any MediaPlayer, if i load the WAV into Audacity, it shows only 10 Minutes, too.

EDIT:

Solution is SOX. Simply convert wav with SOX to w64.

tebasuna51 · 15th October 2023, 02:52

Quote:

Originally Posted by -QfG-

if i convert this wav with FFMPEG to W64:...

Try with:

ffmpeg -ignore_length true -i your.wav output.w64

Without that parameter the output is truncated to the length in the wav header never greater than 4 GB (a field with only 16 bits)

-QfG- · 16th October 2023, 18:26

This works great, thanks, but now i have a "new" Problem^^

If i will demux the Multichannel WAV into MONO WAVs, FFMPEG aborted by 10 Channel (5.1.4) and 12 Channel (7.1.4) files. No error Messages or somethings. 16 Channels (9.1.6) and 8 Channels (5.1.2) works fine.

Code:

"!FFMPEGpath!" -y -threads auto -vsync drop -i "!AVSFILE!.avs" -strict experimental -loglevel error -stats -filter_complex "[0:a]channelmap=0[FL];[0:a]channelmap=1[FR];[0:a]channelmap=2[FC];[0:a]channelmap=3[LFE];[0:a]channelmap=4[SL];[0:a]channelmap=5[SR];[0:a]channelmap=6[BL];[0:a]channelmap=7[BR];[0:a]channelmap=8[WL];[0:a]channelmap=9[WR];[0:a]channelmap=10[TFL];[0:a]channelmap=11[TFR];[0:a]channelmap=12[TSL];[0:a]channelmap=13[TSR];[0:a]channelmap=14[TBL];[0:a]channelmap=15[TBR]" -map "[FL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.L.wav" -map "[FR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.R.wav" -map "[FC]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.C.wav" -map "[LFE]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.LFE.wav" -map "[SL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SL.wav" -map "[SR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SR.wav" -map "[BL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.BL.wav" -map "[BR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.BR.wav" -map "[WL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.WL.wav" -map "[WR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.WR.wav" -map "[TFL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FL.wav" -map "[TFR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FR.wav" -map "[TSL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_SL.wav" -map "[TSR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_SR.wav" -map "[TBL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BL.wav" -map "[TBR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BR.wav"
"!FFMPEGpath!" -y -threads auto -vsync drop -i "!AVSFILE!.avs" -strict experimental -loglevel error -stats -filter_complex "[0:a]channelmap=0[FL];[0:a]channelmap=1[FR];[0:a]channelmap=2[FC];[0:a]channelmap=3[LFE];[0:a]channelmap=4[SL];[0:a]channelmap=5[SR];[0:a]channelmap=6[BL];[0:a]channelmap=7[BR];[0:a]channelmap=8[TFL];[0:a]channelmap=9[TFR];[0:a]channelmap=10[TBL];[0:a]channelmap=11[TBR]" -map "[FL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.L.wav" -map "[FR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.R.wav" -map "[FC]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.C.wav" -map "[LFE]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.LFE.wav" -map "[SL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SL.wav" -map "[SR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SR.wav" -map "[BL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.BL.wav" -map "[BR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.BR.wav" -map "[TFL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FL.wav" -map "[TFR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FR.wav" -map "[TBL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BL.wav" -map "[TBR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BR.wav"
"!FFMPEGpath!" -y -threads auto -vsync drop -i "%~1" -strict experimental -loglevel error -stats -filter_complex "[0:a]channelmap=0[FL];[0:a]channelmap=1[FR];[0:a]channelmap=2[FC];[0:a]channelmap=3[LFE];[0:a]channelmap=4[SL];[0:a]channelmap=5[SR];[0:a]channelmap=6[TFL];[0:a]channelmap=7[TFR];[0:a]channelmap=8[TBL];[0:a]channelmap=9[TBR]" -map "[FL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.L.wav" -map "[FR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.R.wav" -map "[FC]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.C.wav" -map "[LFE]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.LFE.wav" -map "[SL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SL.wav" -map "[SR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SR.wav" -map "[TFL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FL.wav" -map "[TFR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_FR.wav" -map "[TBL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BL.wav" -map "[TBR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_BR.wav"
"!FFMPEGpath!" -y -threads auto -vsync drop -i "!AVSFILE!.avs" -strict experimental -loglevel error -stats -filter_complex "[0:a]channelmap=0[FL];[0:a]channelmap=1[FR];[0:a]channelmap=2[FC];[0:a]channelmap=3[LFE];[0:a]channelmap=4[SL];[0:a]channelmap=5[SR];[0:a]channelmap=6[TSL];[0:a]channelmap=7[TSR]" -map "[FL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.L.wav" -map "[FR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.R.wav" -map "[FC]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.C.wav" -map "[LFE]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.LFE.wav" -map "[SL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SL.wav" -map "[SR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.SR.wav" -map "[TSL]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_SL.wav" -map "[TSR]" -c:a pcm_s%WAVBR%le "!OUTPUTFILE!.T_SR.wav"

SOLVED:

Demuxing works if the W64 files have the correct Channel Layout includet:

Code:

if "!C_LAYOUT!"=="9.1.6 [FL][FR][FC][LFE][SL][SR][BL][BR][WL][WR][TFL][TFR][TSL][TSR][TBL][TBR]" set "PAN= -filter_complex "pan^=7.1^+WL^+WR^+TFL^+TFR^+TSL^+TSR^+TBL^+TBR^|FL=c0^|FR=c1^|FC=c2^|LFE=c3^|SL=c4^|SR=c5^|BL=c6^|BR=c7^|WL=c8^|WR=c9^|TFL=c10^|TFR=c11^|TSL=c12^|TSR=c13^|TBL=c14^|TBR=c15^[a^]^" -map ^"^[a^]^""
if "!C_LAYOUT!"=="7.1.4 [FL][FR][FC][LFE][SL][SR][BL][BR][TFL][TFR][TBL][TBR]" set "PAN= -filter_complex "pan^=7.1^+TFL^+TFR^+TBL^+TBR^|FL=c0^|FR=c1^|FC=c2^|LFE=c3^|SL=c4^|SR=c5^|BL=c6^|BR=c7^|TFL=c8^|TFR=c9^|TBL=c10^|TBR=c11^[a^]^" -map ^"^[a^]^""
if "!C_LAYOUT!"=="5.1.4 [FL][FR][FC][LFE][SL][SR][TFL][TFR][TBL][TBR]" set "PAN= -filter_complex "pan^=5.1^(side^)^+TFL^+TFR^+TBL^+TBR^|FL=c0^|FR=c1^|FC=c2^|LFE=c3^|SL=c4^|SR=c5^|TFL=c6^|TFR=c7^|TBL=c8^|TBR=c9^[a^]^" -map ^"^[a^]^""
if "!C_LAYOUT!"=="5.1.2 [FL][FR][FC][LFE][SL][SR][TSL][TSR]" set "PAN= -filter_complex "pan^=5.1^(side^)^+TSL^+TSR^|FL=c0^|FR=c1^|FC=c2^|LFE=c3^|SL=c4^|SR=c5^|TSL=c6^|TSR=c7^[a^]^" -map ^"^[a^]^""

-QfG- · 18th October 2023, 20:44

Another problem. It looks so, that the Avisynth Timestretch Plugin don't work with Atmos w64 files. Timestretch settings will be ignored :/

Code:

# Load Avisynth Filters
AddAutoloadDir("C:\_RIPPING\FS_Audio_Converter\tools\AviSynth\plugins")
# Source Plugin request: lsmashsource.dll
LoadPlugin("C:\_RIPPING\FS_Audio_Converter\tools\LSMASHSource.dll")
# [Source: LWLibavAudioSource - Stream Index -1]
LWLibavAudioSource("D:\TEST_EAC3ATMOS_3min_[9.1.6].w64", drc_scale=0.0, cache=false, stream_index=-1)
# [DSP: TimeStretch - [Slowdown] 25.00 to 23.976]
ConvertAudioToFloat()
TimeStretch(pitch=95.904)

23rd August 2023, 13:18	#53 \| Link
tebasuna51 Moderator Join Date: Feb 2005 Location: Spain Posts: 6,915	Sorry, I don't know how. __________________ BeHappy, AviSynth audio transcoder.

19th March 2023, 20:34	#42 \| Link
Balling Registered User Join Date: Feb 2020 Posts: 541	Full command "C:\Program Files\Dolby\Dolby Reference Player\gst-launch-1.0.exe" --gst-plugin-path "C:\Program Files\Dolby\Dolby Reference Player/gst-plugins" filesrc location=C:\\tmp\\atmos\\gapless.eac3 ! dlbac3parse ! dlbaudiodecbin ac3dec-drop-delay=true ac3dec-drc-suppress=true out-ch-config=13 ! audio/x-raw,format=S32LE ! wavenc ! filesink location=Z:\zE.L111.wav

23rd March 2023, 17:15	#44 \| Link
Balling Registered User Join Date: Feb 2020 Posts: 541	drc-suppress should be only for EAC3/AC3/AC4, THD has drc off by default. Yes, only int 24 inside int 32 is supported.