600D Audio TEST release - 2.3 based

[miyake]

hmm, qfpost issue again? The file is broken....

Anyway, I think the best way to talk about WAV stereo  is in the WAV forum . 
And I don't wanna import WAV related code in 600D audio repos. If I wrote codes , I will make a patch like wav queue.
It's will make a  confusing to check the codes when mergeing.

[1%]

Its weird, every time I upload something there and someone in another country goes to download it, there are problems. Its still showing as being there for me... is it corrupted. I have to find somewhere else to post.

You don't have to put codes in repo, just asking about stereo vs mono. Anyways, I think the problem was from HPF filter.

[ilguercio]

Use dropbox

[1%]

I should, but I don't like, or rather don't trust, downloading apps for "cloud" services.

http://www26.zippyshare.com/v/72980565/file.html

[miyake]

1%
I just found zippyshare is able to hear the sound directoly on the site.
And hear it, but I don't know what point is wrong you are thougt.

[1%]

This one was stereo, I think problem was HPF but not with code, with HPF itself.

I fixed my beachtek and did some more testing. Monitor sound seems off from sound recorded, i dunno what it is.

Wav recorded along with movie follows audio overrides. Wav recorded through tape recorder function alters volume levels at start.

I messed with some of the wavs produced and I get pretty clean sound if I run a few HPF/LPF filters, hiss goes away.

*Now sound recorder only raises volume the first time (per reboot) its used.

https://bitbucket.org/OtherOnePercent/tragic-lantern/changeset/124d2510cb34cde459362cf3409acf429e112319


So what sound values work for you... internal mic is different from external.

I'm getting decent results from [-2, 0, +6]
With boost from 20-30db.

I think - a little too quiet, 0 is decent 6+ you get the hiss.
Whats better - 0, 25db
                      6, 20db
                      -, 30db?, 25db?

[miyake]

hmm, I still not understanding what you are thinking a problem.
 -Are you investigating a hiss noise?
 -What is problem with HPF ?

I already investigating hiss noise.
http://chirari.ddo.jp/ml/?p=934

Amplifer generated noise is really small. It's depend on which mic we will use.
So I think, We don't need to find best setting for end users.
May be forum users will find best parameters for it.

However , If HPF has some bugs, We must solve it. So please describe detail of HPF issue.

[1%]

When I turn on HPF and record audio it causes ocassional distortion, especially of higher peaks. When I turn it off, distortion is gone.

So I just run a HPF in post and that takes down the bass and takes out some of the hiss. DC offset caused no problems, its literally just DC offset.

When I plug in the beachtek and turn its volume down to 0 I still get hiss. If I adjust analog gain/boost so I barely hear it, I get better sound when I turn the volume back up.

I started recording stuff and seeing how it sounds w/ external XLR mic. I'll need to record a bunch of VO/narration soon and want to be ready. I also need samples for foley so trying to record quieter things which its harder to avoid hiss noise, etc. I can't use PC for this because of the fans, etc.


From charts, no way to escape it. It has 3 bands, 1khz, 3khz, 5khz. I did not have luck recording empty channel and inverting it for some reason but I need to try in real editor and not just audacity.

18db Boost, -12gain looks good... what about 18db boost and 0, -3, +6 or same results.

[miyake]

I got it.
First I will try to reproduce it.

Can I confirm one thing?  Your audio recording means stereo WAV recording, right?

MSG added
Some problem fixed, and pushed your repo. But I can't reproduce it.
I'm testing 1KHz sound generated from my PC and insert mic input on my 600D.
Then testing HPF on/off but it has no problem. actually, I got distortion when input volume really high(red zone).
But I can't find any difference of HPF on/off.....

I think, beachtech input is already had a distortion(check your volume settings on your beachtech.),,,, Just my guess.

[1%]

I think, beachtech input is already had a distortion(check your volume settings on your beachtech.),,,, Just my guess.

Or maybe stereo needs bigger buffer and its from that? I'm testing more to see what happens. Interleave mode on asif recorded wav to video once so I went back.


Trying:

TASK_CREATE( "audio_level_task", compute_audio_level_task, 0, 0x17, 0x1000 );


audio_override_task already at x17

[miyake]

I'm tested with your tragic-lantern binary.
So I also used your stereo codes. And I can't find problems....

[1%]

I'll keep testing... meters better at 0x17 or 0x18? Override task already at 17.

Do you notice it change volume when starting sound recorder first time? It doesn't start running override until the 2nd wav is recorded. But works first time for movie wav for some reason. I added some more override audio . beep.c but it still doesn't override first time.

[miyake]

meter and audio_level tasks are 18 is best. and override is 17.

Oh, tragic-lantern has some limitations? If so, my repo has same issue.
Then we should use override_audio_task(). Then it func will override when started recording.
I just  inspiration hit me.

[1%]

Yes, same problem for your repo. My fix only fixed it for second time recorder starts. Try to record with sound recorder and listen with monitor.  For recording wav with every movie no problem.

Here is a binary from the new compiler. Any faster? Does interleave and sample rate 48000,6 affect high BR + audio?
https://bitbucket.org/OtherOnePercent/tragic-lantern/downloads/autoexec.bin.Speedups.Interleave

[miyake]

Do you notice it change volume when starting sound recorder first time? It doesn't start running override until the 2nd wav is recorded. But works first time for movie wav for some reason. I added some more override audio . beep.c but it still doesn't override first time.

hmm, I can't confirm this problem on my repo. Volume setting is correctly overridden.
Let me clarify your confirm method.
1: Turn on the camera
2: record wav  <--- can't override

Is this a correct?

[1%]

Record first wav or sometimes first movie wav. Audio level heard from headphone goes up. Record second wav or movie wav and volume is at what it supposed to be.

Probably more beep.c than audio.c thing. Only happens on first boot, first time recording. Like audio_configure never ran or was blocked. I didn't try to switch movie mode and back though.

I added  audio_configure(1); to wav_record and wav_recordsmall but it only helps second time.

I'm not sure if the actual volume recorded changes or if just output volume because of canon functions in beep.c but it sounds like boost or analog gain was increased at start. I.e. Start at +6db, +20db and then sounds like +24db, +20db and doesn't change until recording started a 2nd time.

[miyake]

hmmm, I used your binary. But it's correctly override at all times.
I checked volume and watch meters.

The Canon volume is 1(most small).
Then I changing our gain+boost . It's correctly override volumes at 1st time.

[1%]

It just did it with a plug in mic amp... but hiss at 0, off is very good. I recorded wavs, have to look in them and upload. I have external mic selected not auto in/ext, same for you. Also should add back L-ext/R-int , etc.

http://www.magiclantern.fm/forum/index.php?topic=2650.msg10876;topicseen#msg10876

[miyake]

https://bitbucket.org/miyake_t/magic-lantern600daudio

f7924146caf9
Now we don't need to use override_audio_task()!!!!!,It means audio_configure() is enough faster!
Then record start/stop has no delay to override!!!

[1%]

Now overriding is quick but first start problem isn't gone.

SND_0001_FirstStart.WAV

http://www3.zippyshare.com/v/91530107/file.html

SND_0002_SecondStart.WAV

http://www3.zippyshare.com/v/71965814/file.html

http://chirari.ddo.jp/pub/betauploader didn't let me upload wavs still, only bin.

[miyake]

but first start problem isn't gone.

I can't reproduce it.......

[1%]

I took out some audio_configure(1) and now problem only happen if you boot the camera and record new wav with sound recorder first. If you record movie wav first it doesn't happen. But also recorder doesn't re-configure after first record and keeps volume high. Its something that happens first time you press "record new audio clip".

[miyake]

I'm testing only WAV recording.
Anyway, now I can't compile your repo. Are you using gcc 4.7 linaro?
Give me your newest bin.

[1%]

Yes I switched, new compiler makes more optimized bins and performance is slightly better.

https://bitbucket.org/OtherOnePercent/tragic-lantern/downloads/autoexec.bin.wavrec


This is newest, now you can hear it try to do what sound recorder does even when starting recording from rec button (it gets override right away). Something happens first time recording is started but doesn't happen subsequent times.

I experimented a little more. First time StartASIFDMAADC or while that is starting the volume goes up. I did this and it overrides a second after. Maybe I'll shrink msleep amount but doesn't seem like real fix? Every time audio configure there is a pop. Like at camera start there are 3 configures if everything is on.

Code: [Select]

wav_ibuf = 0;
    StartASIFDMAADC(buf1, WAV_BUF_SIZE, buf2, WAV_BUF_SIZE, asif_rec_continue_cbr, 0);
msleep(1000);
audio_configure(1);
   

msleep(100); fixes it but then it pops twice right at start, I guess not so bad and better than alternative.


/OT:

After playing with stand alone pre-amp I think I need something like this for beachtek. Juicedlink wants too much, ($2-300USD)

http://www.ebay.com/itm/New-Nady-DMP-2-Battery-Powered-Dual-Microphone-Preamplifier-3428-/380443876089?pt=US_Pro_Audio_Microphones&hash=item589438daf9

[miyake]

OK! catch it!

audio05 is HPF on and just startup
audio08 is just recording.

Code: [Select]

$ diff /cygdrive/i/ML/audio05.LOG /cygdrive/i/ML/audio08.LOG
18,19c18,19
< ML_MIC_IN_VOL 10
< ML_MIC_BOOST_VOL1 10
---
> ML_MIC_IN_VOL 37
> ML_MIC_BOOST_VOL1 20
34c34
< ML_FILTER_EN 03
---
> ML_FILTER_EN 83
36c36
< ML_DVOL_CTL_FUNC_EN 00
---
> ML_DVOL_CTL_FUNC_EN 2e
41,42c41,42
< ML_EQ_GAIN_BRAND0 e7
< ML_EQ_GAIN_BRAND1 e7
---
> ML_EQ_GAIN_BRAND0 dd
> ML_EQ_GAIN_BRAND1 dd
46,54c46,54
< ML_HPF2_CUTOFF 07
< ML_EQBRAND0_F0L 00
< ML_EQBRAND0_F0H 00
< ML_EQBRAND0_F1L 00
< ML_EQBRAND0_F1H 00
< ML_EQBRAND1_F0L 00
< ML_EQBRAND1_F0H 00
< ML_EQBRAND1_F1L 00
< ML_EQBRAND1_F1H 00
---
> ML_HPF2_CUTOFF 00
> ML_EQBRAND0_F0L 0d
> ML_EQBRAND0_F0H db
> ML_EQBRAND0_F1L 0c
> ML_EQBRAND0_F1H 1b
> ML_EQBRAND1_F0L 6d
> ML_EQBRAND1_F0H 08
> ML_EQBRAND1_F1L 41
> ML_EQBRAND1_F1H ed

Override failed. However, It's late night , I will write code tomorrow. Good night!