ffdshow DTS decoder broken?

[tebasuna51]

Is a audio decoder bug, but maybe here can be read by ffdshow developers.

From sourceforge.net ffdshow-alpha (ffdshow-20041012) to last ffdshow-tryout ffdshow_rev514_20061109_clsid, always I have the following problem.

Decoding a 44100 Hz dts file in GraphEdit with:
Source.dts -> ffdshow Audio Decoder -> WAV Dest -> File Writer
the resultant wav is 2.2% short than original dts or the wav obtained decoding the source with Tranzcode, Foobar2k or NicDtsSource-AviSynth.

Analyzing the wav I found an artifact every 1.06 sec (aprox.) losing 23.22 ms. like show the image:
http://img329.imageshack.us/img329/954/ffdshowdtsaf2.png (http://imageshack.us)
Can anybody confirm this bug?

Edit: Because the source is a 44100 Hz and 1024 samples for DTS frame (0,0232199 sec./frame), seems 2 frames are bad decoded and the third is lose.

Edit 2: With 48 KHz dts source work fine.

Edit 3: The image is a fragment of http://www.sr.se/laddahem/MultiKanal/Dts/SURROUNDTEST_011212.zip

Is a wavdts but for my test I used the extracted dts.

There are more free samples wavdts 44100Hz at:
http://www.sr.se/multikanal/english/e_index.stm

[Rockaria]

Well, I tried my best in my systems but failed to identify to confirm the stated bug.

1. with 44.1k 5.1ch dts files, I compared the durations on foobar2k & ffdshow with both Media Player Classic & graphedit
. the MPC's duration displays around 10% shorter(header identified as 48k?) but the play times are statistically identical(less than 0.01%diff)
. graphedit displays no progress & duration but the play times are identical
2. dumped the wavs from ffdshow & foobar2k finding no difference in the images with audacity. the lenths also are statistically identical(far below 2.2%)
3. downloaded the linked dtswav and played on both foobar2k & ffdshow also finding no noticeable difference.
. MPC's displayed duration was 2.06(126sec) and foobar2k's was 2.07(127sec), truncated display, actual playing times are again,, identical.

It was with the ffdshow version aug 21 2006 with all DSPs off, not an audible artefact to me.

[tebasuna51]

Comparing with Ac3 Filter v1.11, same graph (input, wav dest, file writer) changing only the decoder.

ffdshow_rev514_20061109_clsid without DSP's, DRC, ...
Test.dts -> ffdshow Audio Decoder -> WAV Dest -> File Writer (ffds.wav)

Test.dts -> Ac3 Filter v1.11 -> WAV Dest -> File Writer (ac3f.wav)

ffds.wav 65568814 bytes 2:3.9017
ac3f.wav 67031108 bytes 2:6.6649

- MPC show 1:50 but seems play correctly

[Rockaria]

Interesting... The only difference seems to be the systems involved.

BTW, the MPC I mentioned is the FFDShow activated as the DTSWAV/DTS decoder.

Maybe some other users can input more details.