similar to: near and talk suppressed

Jean Marc & list, Following up on this- there seems to be an issue with this calculation in mdf.c (Smooth echo energy estimate over time) Pey += Eh*Yh; Pyy += Yh*Yh; it goes too large for floating point, which seems to be caused by extremely large values in st->Rf[j], st->Yf[j] and smaller values in st->Eh[j], st->Yh[j]- things were relatively quiet

> I am writing to gain a better understanding of the limitations of speex echo > cancellation, esp. with respect to the fixed point implementation. > If these limitations have been documented elsewhere already, please let me > know! Nothing officially documented, sorry. > I observe experimentally that when one or both of the echo or ref data for > speex_echo_cancel() have

Did a bit of tuning in SVN. Could you try that? What frame size, filter length, sampling rate are you using? All the process you describe is actually the tip of the iceberg I think. There is something making the adaptation diverge, which causes Eh and Yh to be large and so on. The part I changed is probably the one causing divergence. If it doesn't work, reducing the coefficients even more

Hi I just upgraded to http://svn.xiph.org/trunk/speex r10357 and got this build warning: alfredh@io:$ make -s mdf.o libspeex/mdf.c: In function 'speex_echo_cancel': libspeex/mdf.c:321: warning: statement with no effect libspeex/mdf.c:317: warning: `adapt_rate' might be used uninitialized in this function Is this intentional? In any case here is a simple fix: Index:

Hello Jean-Marc, When you count the adaptation step size in case where the filter has already had minimal adaptation, you use the following expression: r = (0.7*r + 0.3*15*RER*e)/e*(power[i] + 10) = (0.7*leak_estimate*Yf[i] + 0.3*15*RER*(Rf[i] + 1))/((Rf[i] + 1)*(power[i] + 10)). Why do we need this weighted sum and the component 0.3*15*RER*(Rf[i] + 1)? Why use the correlation-based RER if

Hi Ted, Thanks a lot for this analysis. > In FLOAT_DIVU() it hangs at the following: > while (a.m >= b.m) > { > e++; > a.m >>= 1; > } > for the case where a and b are both zero (yes, division by zero). > This happens from mdf.c: True, that needs to be fixed even after I fix the rest. > leak_estimate =

Thanks for your prompt and helpful reply. In FLOAT_DIVU() it hangs at the following: while (a.m >= b.m) { e++; a.m >>= 1; } for the case where a and b are both zero (yes, division by zero). This happens from mdf.c: leak_estimate = FLOAT_EXTRACT16(FLOAT_SHL(FLOAT_DIVU(st->Pey, st->Pyy),14)); where st->Pey and st->Pyy are both zero, which happens for the

Hi, I've added visualization of the speex noise cancellation to my program. I did this by taking st->noise[] and st->ps[], scale both by 1.0/(st->ps_size * 32768.0) (to get a value between 0.0 and 1.0), and then draw them as a realtime lineplot. This works well, and my users like being able to see roughly what frequency bands they have noise in and compare it to their input

Jean-Marc, I have traced the second infinite loop further. When st->adapted becomes true (mdf.c line 623), the first Yf[i] value is 4, the leak_estimate is 0xd4e, the resulting r is 3. The first value in st->Rf is 0, so e is 1, and r is set to e>>1, or 0. A little later there is a divide by r, and there is the hang. It seems that the 0 in Rf[0] is the problem, but I am not

Dear list members, I am working with a Cox ph model for the duration of unemployment. The event of interest in my analysis is getting employed. I have various background variables explaining this event: age, sex, education etc. I have multiple unemployment spells per person. I use a model with person-specific frailty terms in order to take into account the correlation of spells by the same

Dear R users, I would like to know if there is a package in R that estimates multinomial logit models (without using surrogate Poisson models) -or even better- nested multinomial logit models? I am using the Windows version of R (Windows 95). To my understanding, the S package called nnet does that, but it seems not to work in R. Thanking in advance for any help, Marjo Pyy-Martikainen

> When you count the adaptation step size in case where the filter has already had > minimal adaptation, you use the following expression: > > r = (0.7*r + 0.3*15*RER*e)/e*(power[i] + 10) = (0.7*leak_estimate*Yf[i] + > 0.3*15*RER*(Rf[i] + 1))/((Rf[i] + 1)*(power[i] + 10)). Actually, the QCONST16(.3,15) doesn't mean (.3*15), but the value .3 in Q15 format (fixed-point). >

Hi: I'm using speex to perform echo cancellation in Windows. I'm aware of the problem about out of sync clocks in record and play sample rates in usual sound cards . In order to have an idea of how good is my echo cancelation working I would like to know if there is any #define thing i can pass to speex_echo_ctl to get the actual level of echo cancellation. If not, how can i extract that

Hi, I've attached a patch against SVN r14645 which adds GCC visibility information to all symbols exported from libspeex.so and libspeexdsp.so. It includes a configure.ac change to test that both the compiler flags and __attribute__((visibility)) works, and if so will #define EXPORT __attribute__((visibility("default"))) and if not #define EXPORT I've attached a diff output

Hi, I've attached a patch against SVN r14645 which adds GCC visibility information to all symbols exported from libspeex.so and libspeexdsp.so. It includes a configure.ac change to test that both the compiler flags and __attribute__((visibility)) works, and if so will #define EXPORT __attribute__((visibility("default"))) and if not #define EXPORT I've attached a diff output

CentOS 7.9 system. My 2 servers at home have been spewing this error message every few minutes to /var/log/messages. But there's no preceding message to tell me what's being suppressed. From googling for this, I understand that the failure to log the real message is a bug in the v3 series kernels and v4 fixes that. 2 servers at the office and another on a VPS with 7.9 do NOT issue

http://bugzilla.mindrot.org/show_bug.cgi?id=1221 Summary: Banner only suppressed at log level = QUIET (used to be at log level < INFO) Product: Portable OpenSSH Version: 4.3p2 Platform: All OS/Version: All Status: NEW Severity: minor Priority: P2 Component: ssh AssignedTo:

After some more analysis of my "dropped fragment" problem, things look like this: Cisco 7940 phone -- RTP --> chan_oh323 --> Asterisk (running, eg., VoiceMailMain) That RTP connection was negotiated via H.323 on a third machine running Cisco CallManager 3.2, but this part should not be relevant. Connections work fine, with one

The immediate reason for this patch is to reduce the amount of debugging in virt-v2v with using the virt-v2v -v option (because this implies running nbdkit in verbose mode too). Most of the messages are datapath ones about pread/pwrite requests, and in fact as we've added more filters on top of nbdkit these messages have got more and more verbose. However they are not particularly