[FFmpeg-devel] [PATCH 9/9] avformat/wavdec: Test s337m

[Tomas Härdin]

mån 2020-01-13 klockan 16:10 +0000 skrev Gaullier Nicolas:
> > > 
> > > +FATE_SAMPLES_AUDIO-$(call DEMDEMDEC, WAV, S337M, DOLBY_E) += 
> > > +fate-dolby-e-wav
> > > +fate-dolby-e-wav: CMD = dolbye2pcm16 -i 
> > > +$(TARGET_SAMPLES)/dolby_e/512.wav
> > > +fate-dolby-e-wav: CMP = oneoff
> > > +fate-dolby-e-wav: REF = $(SAMPLES)/dolby_e/512.wav.pcm
> > > +
> > >  FATE_SAMPLES_AUDIO-$(call DEMDEC, DSS, DSS_SP) += fate-dss-lp 
> > > fate-dss-sp
> > >  fate-dss-lp: CMD = framecrc -i $(TARGET_SAMPLES)/dss/lp.dss -frames 
> > > 30
> > >  fate-dss-sp: CMD = framecrc -i $(TARGET_SAMPLES)/dss/sp.dss -frames 
> > > 30
> > > 
> > This is missing some kind of hash check on the demuxed data
> 
> The "oneoff" tests consists in checking the maximum difference between the raw pcm output samples, it must be 0 or 1 max. This test is done in 16-bit truncated output of the decoded stream.
> It raises an error too if the duration does not strictly match. I found it appropriate (a strict hash on decoded samples may also break with the many floats of the DolbyE decoder).
> My idea was to keep a single test for both "wav demux" and "5.1+2" decode.
> Do you think there should be an additional test focused on demuxed data ?

I don't mean has the decoded data, but rather hash the demuxed packets
before they're decoded.

/Tomas