Brad 2020 #1
Brad 2020 #1
Conversation
| @@ -1111,7 +1116,7 @@ void MainFrame::OnTimer(wxTimerEvent &evt) | |||
| m_textSync->SetLabel("Modem"); | |||
| } | |||
| g_prev_State = g_State; | |||
| if (g_mode == FREEDV_MODE_700D) { | |||
| if ((g_mode == FREEDV_MODE_700D) || (g_mode == FREEDV_MODE_2020)){ | |||
There was a problem hiding this comment.
Good thinking! Hhowever the interleaver (at least past one frame) seems really used, so I doubt it will be used for 2020
There was a problem hiding this comment.
The interleaver might be good for broadcasting.
…Frame::designAnEQFilter to use this, but using stubbed freedv_get_speech_sample_rate.
…current speech sample rate. Replaced instances of FS when used as voice sample rate with freedv_get_speech_sample_rate. This is a pre-test, and testing might show instances of FS being replaced when it shouldn't, and vice versa.
…etting argc to wrong value. In CMakeLists.txt, added lpcnetfreedv to target_link_libraries.
| wxGetApp().m_codec2LPCPostFilterBeta, | ||
| wxGetApp().m_codec2LPCPostFilterGamma); | ||
| } | ||
|
|
||
| // Init Speex pre-processor states |
There was a problem hiding this comment.
@ac0zj How about this Brad? I thought this was perhaps easier than adding (yet another) API function.
There was a problem hiding this comment.
I think that's fine - simple and makes sense!
…t to file. There were/are two 2020 integration issues, one having to do with a fifo not being big enough, the other with pulling too much data out of a sound card fifo. Right now both issues are made to work with hacks - look for bvs in the code. I still need to figure out why the size changes are needed and how those values should be derived, then I'll remove the hacks.
This reverts commit 1d38319.
…mp patch. Comments in code explain what the problem is. Still working on solution to this issue.
…IFO_SIZE to 440 to support 2020.
… to latest stable, 3.0.4.
|
@ac0zj or @hobbes1069 could you pls:
|
Ok, I want to be carefull... This is Pull #1 but mine is Pull #9. Do I need to merge my pull request and then fix Travis CI? |
|
@hobbes1069 no I think that would be too big a step right now. I just need the build instructions pls so I can start testing FreeDV 2020 in this branch ASAP. |
|
@ac0zj I've been working up the Windows side and have the problem traced to the NN decoder drowe67/LPCNet#7 (comment) I have a few local Hams set up to try the Linux version this weekend. I'd like to see how the "human factors" play out in a PTT QSO, in particular sync time. |
|
David,
Sounds good. I got a bit sidetracked with another radio-type project.
I have a co-worker in New Jersey who is interested in FreeDV, so I'll see
if I can get him set up. Otherwise I can try a contact with other U.S.
FreeDV followers.
Cheers,
Brad
…On Fri, May 10, 2019 at 2:52 PM drowe67 ***@***.***> wrote:
@ac0zj <https://github.com/ac0zj> I've been working up the Windows side
and have the problem traced to the NN decoder drowe67/LPCNet#7 (comment)
<drowe67/LPCNet#7 (comment)>
I have a few local Hams set up to try the Linux version this weekend. I'd
like to see how the "human factors" play out in a PTT QSO, in particular
sync time.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/ALUMVP44GAXWWAWRMTRWY2TPUXN7XANCNFSM4HAU3A4Q>
.
|
… We write a new frame of output modem samples as soon as there is room.
|
|
||
| int nsam_in_48 = g_soundCard2SampleRate * freedv_get_n_speech_samples(g_pfreedv)/freedv_get_speech_sample_rate(g_pfreedv); | ||
| assert(nsam_in_48 < 10*N48); | ||
| while((unsigned)codec2_fifo_free(cbData->outfifo1) >= nsam_one_modem_frame) { | ||
|
|
There was a problem hiding this comment.
@ac0zj I think this should be:
while((unsigned)codec2_fifo_free(cbData->outfifo1) >= nsam_one_modem_frame)
to ensure output fifo stays full?
Perhaps we can now try smaller FIFO sizes?
If you switch on Options - txrxDumpFifoState it will print out a bunch of info and you can see it waiting until there is just enough room for a new buffer of modem samples before entering the while loop.
|
I agree - ugh, it should be " >= ", not "<". Yes, I bet we can go back to
smaller FIFO. And I bet it will work well!!
…On Fri, May 24, 2019 at 4:12 PM drowe67 ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In src/fdmdv2_main.cpp
<#1 (comment)>:
>
+ int nsam_in_48 = g_soundCard2SampleRate * freedv_get_n_speech_samples(g_pfreedv)/freedv_get_speech_sample_rate(g_pfreedv);
+ assert(nsam_in_48 < 10*N48);
while((unsigned)codec2_fifo_free(cbData->outfifo1) >= nsam_one_modem_frame) {
@ac0zj <https://github.com/ac0zj> I think this should be while((unsigned)codec2_fifo_free(cbData->outfifo1)
>= nsam_one_modem_frame) to ensure output fifo stays full?
Perhaps we can now try smaller FIFO sizes?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1?email_source=notifications&email_token=ALUMVPZ3SF574LDOGUAN6F3PXBR6JA5CNFSM4HAU3A42YY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOBZWAXUY#pullrequestreview-241961939>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ALUMVP5LJWUS732VXPB3X6LPXBR6JANCNFSM4HAU3A4Q>
.
|
Hi David, here's the pull request. Brad