[Speex-dev] Decoding only a certain frame results in different values than when decoding the entire file
Hermann Weber
hermie.weber at gmx.de
Thu Dec 22 11:20:29 PST 2011
Hi Lakhdar,
yes, sorry, it was a typo. I am decoding bytes 63 to 124, not 63 to 125.
Yes, Case 1 is good.
Yes, what you said is correct: As soon as I reset the memory state, the
output is not as expected anymore.
In Case 1 (which is good), I don't reset the memory state after decoding
1 frame.
In Case 2 I am not skipping the frame by skipping some bytes.
Instead I only read the bytes 63 to 124 from the encoded file and decode it.
Personally, I do not care if the results are binarily the same, but for
some reason, the output in Case 2 is so that I can not work with it.
I can play it, but processing it further is not possible because the
values are different than what I expected.
In the encoded file (when viewed with WinHex) I see something like a
header. Is it really a header? Does it affect the decoding somehow? I am
asking because when I decode bytes 63 to 124 only, I skip this header of
course, and it might be an explanation why the results are different.
In which way does the previous frame affect the next frame?
Thank you so much for the help.
Greetings,
Hermann
Am 22.12.2011 19:53, schrieb Lakhdar Bourokba:
> HI Hermann,
> I am not sure I understand exactly what you are trying to do.
> Let me try to describe what I understood (I assumed your byte array
> starts @ 1 and not 0 as it is supposed to be in C) :
>
> ##############################
> CASE 1: (output is good)
> ##############################
> - INITILIAZE_DECODER (reset state memory)
> - READ bytes 1 to 124
> - SpxDecode(DinBuf, DoutBuf, unsigned int Dinlen)
> - DECODE bytes 1 to 62
> ---OUTPUT Frame 1----
> -293
> -8234
> 2134
> 17
> - DECODE bytes 63 to 124
> ---OUTPUT Frame 2----
> -9323
> -732
> 189
> 2329
>
> ##############################
> CASE 2: (output is not as expected)
> ##############################
> - INITILIAZE_DECODER (reset state memory)
> - SKIP bytes 1 to 62 (This is where I get confused, do you actually
> skip FRAME 1 ?)
> - READ bytes 63 to 124
> - DECODE bytes 63 to 124 (and not 125 as you mentioned, I think
> it's a typo but I just want to make sure)
> ---OUTPUT Frame 2----
> 732
> 9273
> -799
> -723
>
>
>
> If what is described above is what you are doing, you can not expect
> both outputs to be identical.
> The reason is because the decoder has a different state prior to
> processing FRAME 2.
> - In case 2, the decoder has its state reset to the default state.
> - In case 1, the decoder memory state has been modified by the
> processing of FRAME 1.
>
> If I got it wrong, please provide more details.
> L.
>
>
> On Wed, Dec 21, 2011 at 6:03 PM, Hermann Weber <hermie.weber at gmx.de
> <mailto:hermie.weber at gmx.de>> wrote:
>
> Sorry, it seems I have only replied to Lakhdar, not to the newsgroup.
> Below is my reply to Lakhdar, and I would like to make it more clear
> now, using some pseudo values for simplicity:
>
> I read bytes 1 to 124 from my encoded spx file.
> I decode themt and get the values:
> ---Frame 1----
> -293
> -8234
> 2134
> 17
> ---Frame 2----
> -9323
> -732
> 189
> 2329
>
> Both frames are just perfect as I need them.
>
> But now when I read bytes 63 to 125 from my encoded spx and decode
> it, I
> get the values
> 732
> 9273
> -799
> -723
>
> This is weird because I expected to get the encoded values
> -9323
> -732
> 189
> 2329
>
> I hope I could explain it well.
> Am I missing something here?
>
> Thank you very much.
> Hermann
>
>
>
> Hi Lakhdar,
>
> I am not sure if I do anything wrong with the dec state memory.
> Could you perhaps take a look at what I am (perhaps wrongly) doing?
>
> int __stdcall SpxDecode(unsigned char* DinBuf, float* DoutBuf,
> unsigned
> int Dinlen)
> {
>
> //char *testFile;
> //FILE *ftest;
> //testFile = "test2";
> //ftest = fopen(testFile, "wb");
> //fwrite(inBuf,1,inlen,ftest);
> //fwrite(outBuf, 1, outBufpos, ftest);
> //fclose(ftest);
>
> //take every 62 bytes
> //decode, then write to outbuf
>
> char cbits[MAX_FRAME_BYTES];
> float output[MAX_FRAME_SIZE];
> short out[MAX_FRAME_SIZE];
> unsigned int i,j;
>
> unsigned char* tout; tout = (unsigned char*)&out;
>
> for (i=0; i<(Dinlen/62); i++)
> {
> for (j=0;j<62;j++) cbits[j] = DinBuf[(i*62)+j];
>
> speex_bits_read_from(&bits, cbits, 62);
> speex_decode(decstate, &bits, output);
>
> for (j=0;j<160;j++) DoutBuf[j+(i*160)] = output[j];
> }
>
> return 43;
> }
>
> Btw, just if you wondering what I am doing... I am feeding (62 *
> NumberOfFramesThatIWantToRead) bytes from VB6 and return it as a float
> array.
>
> Greetings,
> Hermann
>
> Am 21.12.2011 15:26, schrieb Lakhdar Bourokba:
> > Hi Hermann,
> > Could it be your decoder state memory that is not longer sync ?
> > When a decoder processes a frame, it uses that frame and its memory
> state (affected by the previous frames) to generate the output
> samples.
> If you skip the previous frames, the decoder state memory will not be
> the same (as when you do full decoding) which causes the output to be
> different.
> > I hope this helps.
> > L.
> >
> >
> > On Wed, Dec 21, 2011 at 5:17 AM, Hermann Weber
> <hermie.weber at gmx.de <mailto:hermie.weber at gmx.de>>
> wrote:
> >
> > Hello!
> >
> > I am still using version 1.0.4.
> >
> > When I have encoded a larger file and then only want to decode
> only one
> > frame from it (for example the 2nd frame), is there anything
> special
> > that I have to keep in mind?
> >
> > I thought that I could simply read the bytes from the encoded
> file for
> > example from the start position of the 2nd frame and then decode
> it. I
> > did that, and I can play it play, but the values in the
> decoded 2nd
> > frame are different than they are when I decode the entire file
> and then
> > look at the values for the 2nd frame.
> > I hope I could explain it well.
> >
> > I noticed that there is a pattern in the beginning of the
> encoded
> file,
> > and I am not sure if I have to take it into account somehow.
> >
> > It would be nice if someone could tell me if something comes to
> his/her
> > mind that I might be doing wrong.
> >
> > Thank you very much.
> >
> > Hermann
> > _______________________________________________
> > Speex-dev mailing list
> > Speex-dev at xiph.org <mailto:Speex-dev at xiph.org>
> > http://lists.xiph.org/mailman/listinfo/speex-dev
> >
> >
>
>
> Hello!
>
> I am still using version 1.0.4.
>
> When I have encoded a larger file and then only want to decode
> only one
> frame from it (for example the 2nd frame), is there anything special
> that I have to keep in mind?
>
> I thought that I could simply read the bytes from the encoded file for
> example from the start position of the 2nd frame and then decode it. I
> did that, and I can play it play, but the values in the decoded 2nd
> frame are different than they are when I decode the entire file
> and then
> look at the values for the 2nd frame.
> I hope I could explain it well.
>
> I noticed that there is a pattern in the beginning of the encoded
> file,
> and I am not sure if I have to take it into account somehow.
>
> It would be nice if someone could tell me if something comes to
> his/her
> mind that I might be doing wrong.
>
> Thank you very much.
>
> Hermann
> _______________________________________________
> Speex-dev mailing list
> Speex-dev at xiph.org <mailto:Speex-dev at xiph.org>
> http://lists.xiph.org/mailman/listinfo/speex-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.xiph.org/pipermail/speex-dev/attachments/20111222/679ba232/attachment-0001.htm
More information about the Speex-dev
mailing list