Timing does not match

[Tim Morton]

I just recently found aeneas and it looks like a great tool to index audio files to the text, but I am having trouble getting accurate results.

For instance, I ran the example,

python -m aeneas.tools.execute_task --example-json

and the sonnet.json times generated do not match the sonnet.json sample found in the tool/res folder. 

Here are the times in csv format:

Correct times found in tools/res,

f000001,0.000,2.680,"1"

f000002,2.680,5.880,"From fairest creatures we desire increase,"

f000003,5.880,8.640,"That thereby beauty's rose might never die,"

f000004,8.640,11.960,"But as the riper should by time decease,"

f000005,11.960,14.440,"His tender heir might bear his memory:"

f000006,14.440,18.520,"But thou contracted to thine own bright eyes,"

f000007,18.520,22.720,"Feed'st thy light's flame with self-substantial fuel,"

f000008,22.720,25.480,"Making a famine where abundance lies,"

f000009,25.480,30.400,"Thy self thy foe, to thy sweet self too cruel:"

f000010,30.400,34.400,"Thou that art now the world's fresh ornament,"

f000011,34.400,36.600,"And only herald to the gaudy spring,"

f000012,36.600,40.640,"Within thine own bud buriest thy content,"

f000013,40.640,43.640,"And tender churl mak'st waste in niggarding:"

f000014,43.640,48.080,"Pity the world, or else this glutton be,"

f000015,48.080,53.240,"To eat the world's due, by the grave and thee."

Wrong times generated by above command,

f000001,0.000,2.880,"1"

f000002,2.880,7.280,"From fairest creatures we desire increase,"

f000003,7.280,11.600,"That thereby beauty's rose might never die,"

f000004,11.600,16.480,"But as the riper should by time decease,"

f000005,16.480,20.000,"His tender heir might bear his memory:"

f000006,20.000,24.480,"But thou contracted to thine own bright eyes,"

f000007,24.480,31.720,"Feed'st thy light's flame with self-substantial fuel,"

f000008,31.720,35.480,"Making a famine where abundance lies,"

f000009,35.480,41.800,"Thy self thy foe, to thy sweet self too cruel:"

f000010,41.800,47.400,"Thou that art now the world's fresh ornament,"

f000011,47.400,51.560,"And only herald to the gaudy spring,"

f000012,51.560,53.200,"Within thine own bud buriest thy content,"

f000013,53.200,53.200,"And tender churl mak'st waste in niggarding:"

f000014,53.200,53.200,"Pity the world, or else this glutton be,"

f000015,53.200,53.240,"To eat the world's due, by the grave and thee."

Notice how the last few lines have the same times.

I ran this on a Windows 7 system with Python 2.7. At first I installed all the files using pip and saw the bad timing, then I installed from the Windows group installer .exe but still had the same wrong times. When the files are submitted to the online version they are correct. 

Is there a setting I am missing? What am I doing wrong?

Thanks

Tim

 

[Alberto Pettarin]

Hi,

are you running Win7 32-bit or 64-bit? Is your python version matching that?

I seem to remember a similar issue surfaced in the past, but I cannot
find it and I do not remember the details right now.

If you run with the -v and/or -l flags, does the verbose log show
something useful?

It sounds like an issue with either ffmpeg or the Python C extensions in
aeneas.

HTH,

AP

> --
> You received this message because you are subscribed to the Google
> Groups "aeneas-forced-alignment" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to aeneas-forced-ali...@googlegroups.com
> <mailto:aeneas-forced-ali...@googlegroups.com>.
> To post to this group, send email to
> aeneas-forc...@googlegroups.com
> <mailto:aeneas-forc...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/aeneas-forced-alignment/848a1fbc-e852-49c8-a5bd-ba2e8a726a13%40googlegroups.com
> <https://groups.google.com/d/msgid/aeneas-forced-alignment/848a1fbc-e852-49c8-a5bd-ba2e8a726a13%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Tim Morton]

I'm on 32-bit Windows 7 and 32-bit Python 2.7.

This report seems to indicate the same issue,

https://github.com/readbeyond/aeneas/issues/190

However, I removed v1.7.3 and installed v1.7.0 and the timings are correct. According to the report above the error occurred even with v1.7.1 so that should narrow it down some.

Also, even with v1.7.0, the cew C plugin will not load even though the cew.pyd file is in the cew folder. Here is the diagnostics report. 

[INFO] ffprobe        OK

[INFO] ffmpeg         OK

[INFO] espeak         OK

[INFO] aeneas.tools   OK

[WARN] shell encoding WARNING

[WARN]   The default input encoding of your shell is not UTF-8

[WARN]   The default output encoding of your shell is not UTF-8

[INFO]   If you plan to use aeneas on the command line,

[INFO]   you might want to 'set PYTHONIOENCODING=UTF-8' in your shell

[INFO] aeneas.cdtw    AVAILABLE

[INFO] aeneas.cmfcc   AVAILABLE

[WARN] aeneas.cew     NOT AVAILABLE

[INFO]   You can still run aeneas but it will be a bit slower

[INFO]   Please refer to the installation documentation for details

[WARN] All required dependencies are met but at least one Python C extension is

not available

Thanks,

Tim

> an email to aeneas-forced-alignment+unsub...@googlegroups.com
> <mailto:aeneas-forced-alignment+unsubscribe@googlegroups.com>.

> To post to this group, send email to

> aeneas-forced-alignment@googlegroups.com
> <mailto:aeneas-forced-alig...@googlegroups.com>.

[Alberto Pettarin]

Right, #190 it was.

So there seems to be a regression from 1.7.0 to 1.7.1, and it affects
64-bit Windows as well, not just 32-bit Windows.

I only own Linux machines and an old Mac Mini, so aeneas is "officially"
supported only on Linux (and, to lesser extent, macOS), I will see if
next week (I am busy at work + traveling these days) I can set a Windows
VM up in VirtualBox and reproduce/diagnose the issue.

AP