eac3to releases

[Columbo]

eac3to build 3.53:

* Fixed dial norm removal for non-Atmos THD. Disallowed removal for Atmos THD.
* Fixed bugs in -edit for AC3.
* Allow processing of a single AC3 frame.
* Fixed mono LPCM in MKV.
* Implemented dial norm removal for DTS-MA audio.
* Fixed demuxing compressed PGS subtitles from MKV.
* Updated the changelog.txt file.

https://rationalqm.us/eac3to/eac3to_3.53.rar

[Columbo]

eac3to build 3.54:

* For bluray/UHD disks, added skip first M2TS (-sf) and skip last M2TS (-sl)
options. These allow for proper handling of Kino-Lorber-like disks and other
disks with a spurious last M2TS.
* Fixed a bug in STN table parsing that caused incorrect marking of hidden
streams in the UI.
* Fixed bugs in detection of src/dest overlap detection. This detection is
intended to prevent accidental overwrite of source files.
* Fixed cases where a source file could be improperly deleted, even if the
command line does not have a src/dest overlap.
* Fixed some cases of improper framerate detection for HEVC streams.

Hat tip: Curly, oniiz86, DelBoy83, GCRaistlin

https://rationalqm.us/eac3to/eac3to_3.54.rar

[Columbo]

Guys, please re-download eac3to 3.54 now if you had gotten it before, as a regression has been fixed. Sorry for the inconvenience.

[Columbo]

One more slipstream fix for eac3to 3.54: Fix wrong chapter name string for the first chapter when skip first is in effect. Please re-download 3.54 to get this fix. Hat tip Island.

[Columbo]

eac3to build 3.55:

* Accept DTS-MA streams without a core. The UI display is still wrong
for now and needs to be fixed, but the streams can be opened and processed.
* In the UI and in the demuxed filenames when using -demux, iso639 names
are now used. For example, instead of English we now have [eng]. This
change allows for mkvtoolnix to autodetect the language from the filename(s).

https://rationalqm.us/eac3to/eac3to_3.55.rar

[Columbo]

Slipstream fix for eac3to 3.55:

* Fixed regression in 3.54 that broke -slowdown.

https://rationalqm.us/eac3to/eac3to_3.55.rar

[Columbo]

eac3to build 3.56:

* Fix crash with the -sf option.
* Fix collision between -sl and -slowdown.
* Add INI file option -trim_last_chapter. Refer to the user text document
for details.
* Update user text document.

https://rationalqm.us/eac3to/eac3to_3.56.rar

Hat tip: Curly, oniiz86, DelBoy83, HPotter

[Columbo]

eac3to v3.58

* Now disables system sleep while processing and reenables it afterwards.
* Fixed handling of custom channel maps.
* Revised reported info relating to dial norm processing to be more consistent.
* Fixed demuxing of HEVC video from MKV.

https://rationalqm.us/eac3to/eac3to_3.58.rar

Hat tip: Curly, Rocky, oniiz86, GCRaistlin

[Baldrick]

I guess it should be https://rationalqm.us/eac3to/eac3to_3.58.rar

[Columbo]

Oops, thank you sir!

[Columbo]

eac3to v3.59

* Fixed some EAC3 7.1 streams detected and shown as 5.1.
* Fixed EAC3 6.1 streams detected and reported as 5.1.
* For EAC3 EX streams, display "EX" only on the core.
* Fixed crash on streams with too many (>256K) leading zeros. Now
reports "The format of the source file could not be detected."
* Detect and report EAC3 Atmos.
* Disable DN for EAC3 Atmos as already done for THD.
* Improved reporting of DN events.
*Changed the format names for DTS-MA and DTS-HR to be consistent
with reporting for other formats.
* Removed dependency on msvcr100.dll (from zlib1.dll).
* Fixed issues arising when directly opening an M2TS file within a bluray structure,
rather than opening the disk.
* Refactoring of eac3to.dpr to improve readability/maintainability.

https://rationalqm.us/eac3to/eac3to_3.59.rar

Hat tip oniiz86, Drommer, GCRaistlin

[Columbo]

eac3to version 3.60

https://rationalqm.us/eac3to/eac3to_3.60.rar

* Fixed failure to detect format for several .ts file edge cases.
* Added a string to the end of the video report line that indicates: SDR/HDR10/HDR10+/HLG/DoVi.
* Fixed PGS sync for bluray/UHD disks. Tested successfully with LITTLE_MONSTERS and MONSTERS_UNIVERSITY.
* Implemented THD gaps correction (uses the DGDemux algorithm).
* thd+ac3 extension now works correctly. If you demux with thd+ac3 (e.g., eac3to g:\rips\DISC 2) 2: output.thd+ac3) then you'll get the .thd, the embedded .ac3, and the .thd+ac3 (all gaps-corrected). To get .thd and/or .ac3 without merging, just specify the individual tracks as usual.
* The thd+ac3 deprecation notice was removed from the notes file.
* Fixed a latent bug in THD parsing that could crash eac3to.
* When -showall is not given and a disk has no playlist longer than 15 min, then an error message
was given saying disk structure could not be found, which is misleading. The error message for
this case was changed to "All Blu-Ray playlists are shorter than 15 minutes. Try using -showall."
* Added info message about Nero decoding of mono AAC with wav output.
* Fixed missing forced subtitles when using the -forced option.

Hat tip: Peter, oniiz86, GCRaistlin, Drommer, Superb, J4ZzY, madshi

[Columbo]

eac3to version 3.61:

https://rationalqm.us/eac3to/eac3to_3.61.rar

* Fixed detection of E-AC3 7.1 in M2TS/MKV.
* Distinguish between 7.1 and 5.1.2 for E-AC3 JOC (Atmos).
* Now print "E-AC3 JOC (Atmos)" instead of "E-AC3 (Atmos).
* Fixed reporting of DTS Master Audio without core in MKV/MKA/M2TS/TS.
* Instead of "no core" print "without core".
* Fixed "eac3to disk" does not show subs, while "eac3to disk 1)" does show subs.
* Sometimes the duration shown for an MPLS is different depending on whether the eac3to
command is "eac3to disk" or "eac3to disk 2)". Now they will differ only if
one or more of the M2TS files has video outside the MPLS window, so you can use the
discrepancy to know that this is the case.

Hat tip oniiz86, Drommer, Paul3

[Columbo]

eac3to 3.62:

* Revised behavior of -showall so it is not needed for some pathological disks.
Playlists will always be shown if they are long enough.
* Fixed bugs in -sf/-sl.
* Detect and report DTS:X streams.
* Invoke ffmpeg to decode E-AC3 7.1 and 5.1.2 streams. Fixes the previous limitation
that 7.1 was decoded as 5.1.
* Correct reporting of dial norm actions for 6.1 EAC3 files with core.

https://rationalqm.us/eac3to/eac3to_3.62.rar

[Columbo]

eac3to 3.63:

* Added support for AC3 streams with bsid 9 and 10.
* Added -numLogFiles=xxx option to set the number of log files to retain.
* In the report lines, DTS L/R is now distinguished from DTS Lt/Rt.
* Added support for AUD-less AVC and HEVC.
* Fixed spurious "The video track has a non-standard framerate." messages.
* Fixed crashing during format detection for some AVC/HEVC streams.
* Fixed spectral "garbage" (banding and haze) when decoding AC3/EAC3.
* Fixed wrong handling of downmix that caused spikes in the spectrogram
when decoding AC3/EAC3.
* Dialnorm removal was changed to run always on all frames and not
skip normalization if the first frame already has -31dB. DN is a per
frame property and it can differ between frames.
* Fixed format detection for some 7.1 JOC streams.
* The -ffmpeg option now supports wav, wavs, w64, rf64, and flac output.
* Documents were updated.

https://rationalqm.us/eac3to/eac3to_3.63.rar

[Columbo]

eac3to 3.64:

* Fixed writing of wrong new frame rate to the output file in some cases when the ES frame rate is missing.
* Updated libflac.dll to version 1.5.0.
* Set 8 threads for FLAC encoding. In one test it reduced the encode time from 47 seconds to 10 seconds!
* Fixed a regression in format detection. Some WAV files were detected as video.
* Fixed a regression in sup file demuxing when the input is multiple concatenated M2TS files.
* Fixed an old bug in demuxing eac3 from M2TS for some streams.

https://rationalqm.us/eac3to/eac3to_3.64.rar

Hat tip: oniiz86, Thunderbolt, kz, tvholic, HPotter, GRTR

[Columbo]

eac3to 3.65:

* Fixed a bug causing crashes with some float WAV files.
* Fixed a crash when demuxing more than 64 tracks at once (e.g. MKVs with
dozens of audio/subtitle language tracks).
* Fixed file creation failure when a track title contains path characters
(e.g. leftover ripper temp paths in MKV track names).
* Fixed overly long track titles that could exceed the Windows path limit and
abort demuxing (title portion of auto-generated filenames is now capped at 128 characters).
* Declared tracks containing no data are now reported (e.g., "Video track N contains no frames." or
"Audio track N contains no data." or "Subtitle track N contains no captions."). Empty files are not created.
* Show "DTS 96/24" instead of "DTS-96/24" and "DTS 96/24 ES" instead of "DTS-ES-96/24" for all DTS 96/24 streams.

https://rationalqm.us/eac3to/eac3to_3.65.rar

Hat tip: GCRaistlin, Drommer, oniiz86

[Columbo]

eac3to 3.66:

* Added -strictpl option: demux only the data inside the playlist's M2TS in/out time windows.
Refer to eac3to_3.66_notes.txt for details.
* eac3to now reports when a playlist demux encounters data outside the playlist time windows.
Refer to eac3to_3.66_notes.txt for details.
* Added -auditpl option: audits a playlist without demuxing. Refer to eac3to_3.66_notes.txt
for details.
* 2D playlists on 3D-capable disks no longer pretend to be 3D. The playlist's own stereoscopic
declaration is now read, so a 2D title neither lists nor demuxes the MVC stream (it remains
selectable by its track number), while true 3D titles are unchanged. A nice side effect:
just listing a disk now shows which titles are 3D and which are 2D.
* Fixed hidden-stream (*) marking on multi-angle playlists: the STN table was read from the wrong
offset when the first play item had multiple angles, so most tracks were wrongly marked as hidden.
* Added -include_hidden option (command line or INI). -demux and wildcard demuxing now skip tracks
marked hidden (*) unless this option is given. Refer to eac3to_3.66_notes.txt for details.
* -demux output files from a Blu-ray/UHD title are now named after the playlist (MPLS) number
instead of the first M2TS number, so titles sharing a first M2TS number no longer overwrite each other.
* Updated the changelog and made eac3to_3.66_notes.txt.

Hat tip: DelBoy83, Curly

https://rationalqm.us/eac3to/eac3to_3.66.rar

[Columbo]

eac3to 4.00:

* Replaced the old avcodec-54/avutil-52 with a modern build (avcodec-62/avutil-60) for AC-3, E-AC-3,
MP1/MP2/MP3, TrueHD, MLP, and DTS. Refer to eac3to_4.00_notes.txt for details.
* E-AC-3 7.1 and Atmos (JOC) streams are decoded natively. The -ffmpeg option and the bundled ffmpeg.exe are gone.
* AC-3 and E-AC-3 are decoded without dynamic range compression, as the -ffmpeg path already did.
* DTS, DTS-HD, DTS Express and core-less DTS-HD are now decoded by avcodec-62 by default. -dcaDec selects
libDcaDec and -arcSoft selects ArcSoft as before; -libav is no longer needed. Lossless DTS-HD output is
unchanged and written at the stream's own bit depth in one pass.
* Fixed AAC files and MKV/MKA tracks shorter than about 40 KB not being detected.
* Fixed AAC tracks in MKV/MKA with channel configurations 11 to 14 (6.1, 7.1, 22.2, 5.1.2) aborting with a libav
error and being listed as 5.1.
* Fixed AAC tracks whose PCE declares a side pair (e.g. 7.0) decoding with the channels mislabelled and out of order.
* Fixed AAC tracks in MKV/MKA files written by ffmpeg, whose frames after the first keep their own PCE, failing to
decode.
* AAC is now decoded by avcodec-62. Nero or Sonic is used only when -nero or -sonic is given.
* HE-AAC tracks are listed with the decoded sample rate and channel count and named "AAC (HE-AAC)"
or "AAC (HE-AAC v2)". Mono HE-AAC is decoded as mono instead of two identical channels.
* -mono now writes the mono channel mask. The WAV header previously claimed a stereo layout for one channel.
* Fixed AAC tracks in MKV/MKA that could not be decoded: HE-AAC with explicit signalling in the codec
private data was read as a stream header, and tracks whose channel layout is given by a program config
element had no layout in their frames. Both now decode and list correctly.
* Fixed MLP channel assignments 18, 19, and 20 (the layout DVD-Audio uses for 5.1) decoding with the centre and
LFE channels swapped against the surrounds. Present since 3.3x.
* A stream whose channel count or sample rate changes mid-way now stops with a message instead of producing
a broken WAV file, as the DTS path already did.
* The channel mask written to WAV, W64, and RF64 files now comes from the decoder's own channel layout. Some
DTS files previously got a mask with more bits set than channels.
* New log lines: "Decoding with libav/ffmpeg (avcodec-62)..." marks the new decoder; "reported a non-native channel layout"
marks a stream whose channel order could not be verified, please report it with the sample; "switched its sample format"
is informational only.
* Fixed AC-3 decoding in normal mode (no -fast) stopping at the first frame with a bad CRC and writing a truncated file.
Such frames are now reported, counted, and decoded anyway.
* In normal mode (no -fast), AC-3 and E-AC-3 frames that fail their CRC check are now reported and replaced
by repeats of the last good block, as in 3.6x. In fast mode nothing is checked and such frames are decoded
as they are, which is how ffmpeg-based players treat them.
* Fixed .wavs, .raw, .pcm, .lpcm, and .agm output losing the delay and every other per-track option
(gain, downmix, resampling, channel map) when a second pass was needed, typically for a 24-bit track that
only holds 16-bit data.
* A file inside a Blu-ray disc folder (e.g. a WAV in the disc root) no longer prints the
"All Blu-Ray playlists are shorter than 15 minutes" message.
* Updated user docs.

https://rationalqm.us/eac3to/eac3to_4.00.rar

Hat tip: interdude, oniiz86, DelBoy83, Curly, madshi, squash

[Columbo]

eac3to 4.01:

https://rationalqm.us/eac3to/eac3to_4.01.rar

* AAC encoding is built in: the libav AAC encoder replaces NeroAacEnc, nothing needs to be installed. Output is
ADTS .aac; .m4a/.mp4 output is not supported. -quality and -NNN work as before. HE-AAC is no longer produced.
Refer to eac3to_4.01_notes.txt for details.
* The Nero, Sonic, ArcSoft, and libDcaDec decoders are removed. -nero, -sonic, -arcSoft, and -dcaDec are accepted and ignored with a notice. libdcadec.dll is gone.
* MKV output now muxes through libavformat instead of the Haali DirectShow muxer, so no DirectShow filters need to be installed. As before, only the video track goes into the MKV; audio, subtitles, and chapters are written to separate files as in earlier versions. H.264, VC-1, and MPEG-2 video are muxed as before, and HEVC is new: 4.00 refused it. The .gaps file is no longer written, and -seekToIFrames is accepted and ignored.
* DTS encoding is built in too: the libav DTS core encoder replaces SurCode. 768 or 1536 kbps (-768/-1536), mono,
stereo, and 5.1; 6.1/7.1 sources are downmixed. Core DTS only.
* AC3 encoding moved from libaften to the libav AC3 encoder; libaften.dll is gone. Same bitrates, defaults, and
layouts; output is not byte-identical to 4.00.
* A delay, gap, or edit on a dual mono (1+1) AC3 track is now padded with copies of the track's first frame instead
of silence, with a warning. The libav AC3 encoder has no dual mono layout.
* Encoding from a WAV file whose channel mask doesn't match its channel count now stops with an error.
* EAC3 encoding is new: any source can target .eac3 (or .ec3/.ddp/.dd+), 5.1 at most, same bitrate list as AC3.
An EAC3 source to .eac3 is still a copy.
* New -allowDrc option (command line or INI): the AC3/EAC3 decoder applies the stream's DRC words, as players do. Off by default, as in 4.00.
* New -applyDialnorm option (command line or INI): decoded AC3/EAC3/TrueHD (and AC3 inside WAV) is attenuated by 31 minus the stream's dialnorm, as a Dolby decoder does. Off by default. An AC3 or EAC3 source going to .ac3 or .eac3, also with -core, is re-encoded instead of copied. Tracks that are only copied, such as TrueHD to .thd, are handled as without the switch, so -keepDialnorm still applies to them; .thd+ac3 output is left unchanged, and -check ignores the switch. Combined with a +/-dB gain, the gain line shows the breakdown: "Applying 2dB gain (8dB requested, -6dB dialnorm)".
* The -test option is removed; it is accepted and ignored with a notice.
* The -r8brain option is removed. It had been disabled for years, and SSRC has always done the resampling;
r8b.dll is gone. -r8brain is accepted and ignored with a notice.
* eac3to no longer touches the Windows registry and needs no third-party software installed.
* Fixed raw (E-)AC3 files with junk before the first frame, or cut mid-frame, failing with "The format of the source
file could not be detected". Detection and decoding now start at the first complete frame.
* Fixed the success sound not playing after a chapters-only demux.
* The caution for a WAV file bigger than 4GB now suggests a .w64 or .rf64 destination.
* Fixed the phase shift option for mono tracks (a number of degrees followed by the degree sign) not being recognized. Angles that are not a multiple of 90 are now applied correctly too.
* Fixed the language names Bokmal, OldProvencal, and Volapuk being shown as garbled text.
* HEVC streams that don't flag themselves as interlaced are now listed as progressive; 4.00 listed some
progressive streams as interlaced at double the frame rate (e.g. 1080i60/1.001).
* Fixed the gain line in the log rounding positive gains to two significant digits (+10.5dB showed as 11dB).
* Updated user docs.

Hat tip: GCRaistlin, Curly, DJ Bobo, HPotter, new_guy, Claude, Thunderbolt, madshi, tebasuna51, oniiz86, Britney

[El Heggunte]

Many thanks! for the overdue updateS.

[Columbo]

You're welcome and thank you. The 64-bit version is on a test release and linux is coming.

------------------
"AI coding agents are machine-shaped machines that transform electricity into confident bugs, at scale, with a summary saying all tests passed."

[Columbo]

eac3to for linux test build:

https://rationalqm.us/eac3to/eac3to_4.02_linux_test.tar.gz

[babiulep]

Hello Columbo!

Thanx for the linux build.
Realy early feedback...

working:
- converting 5.1 DTS-HD to single multi channel wav
- converting 5.1 DTS-HD to ac3 downmixed to stereo (detected clipping and started 2nd pass)

not working (but might be early days...):
- converting 5.1 DTS-HD to aac
(The libav decoder crashed. Aborted at file position 262144)

Greetings!

[Columbo]

Thank you sir for the feedback! Yes, it is the very first test release. I will look into that issue straight away.

[Columbo]

The problem with AAC encoding was fixed. Re-download the test build and update eac3to. You have earned a spot in the contributors list. Thank you!

BTW, if you like, you can follow development more closely here:

https://www.rationalqm.us/board/viewtopic.php?p=25230#p25230

[babiulep]

The problem with AAC encoding was fixed. Re-download the test build and update eac3to. You have earned a spot in the contributors list. Thank you!

BTW, if you like, you can follow development more closely here:

https://www.rationalqm.us/board/viewtopic.php?p=25230#p25230

Yip, it's working! And thanks for the link.

Track 2 is used for destination file "output.aac".
[a02] Extracting audio track number 2...
[a02] Decoding with libav/ffmpeg (avcodec-62)...
[a02] Downmixing multi channel audio to stereo...
[a02] Encoding AAC <256kbps> with libav/ffmpeg (avcodec-62)...
[a02] Creating file "output.aac"...
[a02] Clipping detected, a 2nd pass will be necessary. <WARNING>
[a02] The original audio track has a constant bit depth of 24 bits.
[a02] The processed audio track has a constant bit depth of 64 bits.
[a02] Starting 2nd pass...
[a02] Extracting audio track number 2...
[a02] Decoding with libav/ffmpeg (avcodec-62)...
[a02] Downmixing multi channel audio to stereo...
[a02] Applying -4.75dB gain...
[a02] Encoding AAC <256kbps> with libav/ffmpeg (avcodec-62)...
[a02] Creating file "output.aac"...
[a02] The processed audio track has a constant bit depth of 64 bits.
Video track 1 contains 129796 frames.
eac3to processing took 13 minutes, 30 seconds.
Done

[Columbo]

Hehe, Claude doesn't even have to use a debugger. It's astounding.

[Columbo]

Also see here for some history:

https://www.rationalqm.us/board/viewtopic.php?p=24878#p24878

[NovaHunter47]

Nice to see dial norm removal extended to DTS-MA; disabling it for Atmos THD is the safer call since metadata handling there can get messy. The compressed PGS demux fix from MKV should help a lot with subtitle workflows.