similar to: Noisy sound quality with Blackfin in WB-mode

Hi Bernhard, It seems like I messed up somewhere in the lpc code. However, it seems like your patch is also removing some code, which is probably why you see small differences between your asm version and the fixed point. I don't have my setup working at the moment, but could you test the following patch: Index: lpc_bfin.h ===================================================================

Hello Jean-Marc, I disabled the Blackfin optimization functions one by one and found the error in libspeex/lpc_bfin.h in the autocorrelation computation. Compared to the fixed-point-variant of that function (from lpc.c), the blackfin-code produces correct values except for the zero offset value. Additionally, the loop after the ASM-code assigned values (ac32) wrong to the functions

Hi! It seems that using a different compiler (I assume the development is done with MSC) uncovers various issues with the floating point. Here is the second one I'm seeing - DOMAIN error from sqrt(). In the open_loop_nbest_pitch(), this line: g = DIV32(corr[i-start], 10+SHR32(MULT16_16(spx_sqrt(e0),spx_sqrt(energy[i-start]) ),6)); The values are: e0 is 7.2 i is 142, start is 17

Most asm statements clobber ASTAT bits (shifts, maxes, etc...) but do declare the register as clobbered. Same thing with CC in a few places. Some places make an attempt at clobbering some hardware loop registers, but it's very incomplete compared with how many asm statements actually use hardware loops. Signed-off-by: Mike Frysinger <vapier at gentoo.org> --- libspeex/bfin.h

Hello I'm testing SPEEX on embedded board using ARM7 (Atmel). ARM7 don't have floating point so I'm using FIXED_POINT. Unfortunately the encoding speed is about 5 times slower then necessary for real time. ARM7 is slow for 16/8 bits operations. The sequence: static inline spx_word32_t compute_pitch_error(spx_word16_t *C, spx_word16_t *g, spx_word16_t pitch_control) {

Did you check it against the trunk in SVN? If it's not applied, and you can hook Jean-Marc up with an email address like yours, I'm sure he will get right on it. :) Tom Mihai Balea <mihai@hates.ms> wrote: > > Hi all, > > Did anything happen to this patch? > It seems to me that it fixes a valid issue, but I'm not an expert. > Anyways, I didn't see

There is a problem in speex_preprocess_ctl. Both speech_prob_start and speech_prob_continue are set to 327.67 for all input values except 0 which results in 0. This is in floating point mode. I think the included patch fixes the problem. Mikael -------------- next part -------------- Index: libspeex/preprocess.c =================================================================== ---

> I was wrong, it is the encoder that is not working, and it stopped working > in build 11103. The log message for this build is "another 640 bytes > removed from the encoder state (using the input data instead of copying it > to st->frame/st->inBuf)". Only nb_celp.c/h are changed in this build. What > I am seeing out of the decoder is an extremely low signal

Hi, Ref the disucussion on IRC yesterday; here's a patch which makes a bit more data from the analysis of the preprocessor and the echo canceller available. For the preprocessor: - Size of power spectrum. - Power spectrum and noise estimate of the previous frame. These are given as squared values, so sqrt() to get values in the 0->32767 range. - Current amplification level

From: Thorvald Natvig <slicer at users.sourceforge.net> Some optimizing compilers miscompile the current SSE optimizations when full optimizations are enabled. By using output value pointer instead of a return value, we can bypass this misbehaviour. --- libspeex/resample.c | 8 ++++---- libspeex/resample_sse.h | 24 ++++++++---------------- 2 files changed, 12 insertions(+), 20

Hi Jean-Marc, I could be way off base here, but it seems to me that line 115 in pitch.c in the function "pitch_downsample": x_lp[i] = SHR32(HALF32(HALF32(x[1][(2*i-1)]+x[1][(2*i+1)])+x[1][2*i]), SIG_SHIFT+2); should actually be: x_lp[i] += SHR32(HALF32(HALF32(x[1][(2*i-1)]+x[1][(2*i+1)])+x[1][2*i]), SIG_SHIFT+2); Sorry if I'm totally misreading things and just

--- celt/arch.h | 5 +++++ silk/macros.h | 4 +--- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/celt/arch.h b/celt/arch.h index 9f74ddd..670527b 100644 --- a/celt/arch.h +++ b/celt/arch.h @@ -78,6 +78,11 @@ static OPUS_INLINE void _celt_fatal(const char *str, const char *file, int line) #define UADD32(a,b) ((a)+(b)) #define USUB32(a,b) ((a)-(b)) +/* Set this if opus_int64

Hello, in celt/celt_encoder.c line 369, the 'b' argument to MAX16 can sometimes be greater than what can be represented by a 16bit integer. The default definition of MAX16 is type-less, but I am working on an architecture with hardware support for min/max of 16bit. Changing the default definition to take advantage of this hardware changes the result of that computation. Please consider

Hi, I am using 1.2beta2 on a C5416 doing narrowband in the decode function I am having a problem with the execution of my program vectoring off to random interrupts that are not used. I have tracked down the error to the highpass function called in nb_decode -very close to the end of the function-context: it is called here: ln 1719 for me if (st->highpass_enabled) highpass(out,

Hi, I'm fairly new to Speex, so please forgive me if this question has an obvious answer. I'm currently studying the source code to understand how Speex works and wondered, is there still a floating-point implementation I could look at as the fixed-point stuff makes it a little difficult to see what the code is doing. Regards, Matt.

This was reported by me (and most likely others). Here is the exchange about this. The reason the problem is visible in Delphi is that Borland (C++ and Delphi) initialize the FPU differently from other RTLs. Also, there is another problem that will be visible with Delphi --------------------------------------------------------- ------------- Yes, that's definitely a bug in the vbr code,

From: Jyri Sarha <jsarha at ti.com> I optimized Speex resampler for NEON capable ARM CPUs. The first patch should speed up resampling on any platform that can spare the increased memory usage. It would be nice to have these merged to the master branch. Please let me know if there is anything I can do to help the the merge. The patches have been rebased on top of master branch in

--- configure.ac | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/configure.ac b/configure.ac index f52d2c2..e1a6e9b 100644 --- a/configure.ac +++ b/configure.ac @@ -190,7 +190,7 @@ AC_ARG_ENABLE([rtcd], [enable_rtcd=yes]) AC_ARG_ENABLE([intrinsics], - [AS_HELP_STRING([--disable-intrinsics], [Disable intrinsics optimizations for ARM(float) X86(fixed)])],, +

Hi All, When reading some files we get this error: md5sum: /path/to/file.xml: Structure needs cleaning in /var/log/glusterfs/mnt-sharedfs.log we see these errors: [2013-12-10 08:07:32.256910] W [client-rpc-fops.c:526:client3_3_stat_cbk] 1-testvolume-client-0: remote operation failed: No such file or directory [2013-12-10 08:07:32.257436] W [client-rpc-fops.c:526:client3_3_stat_cbk]

Hi Jean-Marc, Hi Jim, I have also seen some problems with the 1.1.8 release on the C55x. So far I have boiled down the issues to the following: 1) We need our own "fixed_xx.h" header file. I don't know why, and haven't had time to investigate, but there is a definite improvement when I use the attached fixed_c55x.h file which has turned all the maths into inline functions.