phone compress mic signal

[tonko....@gmail.com]

hi, i have used for years dayton mic and calibrated phone mic on samsung s5 and samsung tab s2. it worked perfectly on both devices!!! now i have new phone, samsung a41, that compress mic input signal and i can't use audio tool for noise level measeurements any more. is any solution for this problem? thank you!

[john pf]

Ill try it and let you know.

On Tue., Jul. 21, 2020, 9:42 a.m. , <tonko....@gmail.com> wrote:

hi, i have used for years dayton mic and calibrated phone mic on samsung s5 and samsung tab s2. it worked perfectly on both devices!!! now i have new phone, samsung a41, that compress mic input signal and i can't use audio tool for noise level measeurements any more. is any solution for this problem? thank you!

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/46540c58-b043-4e2c-ae19-44fede6b514ao%40googlegroups.com.

[Julian Bunn]

It's disappointing news if Samsung are no longer respecting the "VOICE_RECOGNITION" Android flag in the A41 - this SDK flag is supposed to guarantee a stream of audio samples from the microphone that have no filters or AGC applied, which is clearly important for AudioTool.

What are the symptoms of the problem on the A41 - how do you see the compression?

Julian

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAL2BMHd5-2qU-HP_mDLvDf5q44bT8%2BrsioEjcY9VQyVpS8BX4g%40mail.gmail.com.

[tonko....@gmail.com]

i first compared sound meter with dayton mic on old S5 phone at different noise levels.source noise varied between 60 and 100 dB. they always had the same measure results.

after that i compared the sound meter with dayton mic on new A41 phone. source noise varied again between 60 and 100 dB. at low levels A41 had higher measure results then sound meter. at 100 dB source level A41 had only measured 90 dB.

the same was when i used A41 phone original mic instead of dayton mic.

[Julian Bunn]

Do you see any difference if you switch to the Front built in mic (via the Menu in AudioTool)? 

Julian

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/b503385c-1e86-4d7d-b113-f27c9a4e960bo%40googlegroups.com.

[tonko....@gmail.com]

with the front mic is the same. it can only measure up to 85 dB. althogh i incresed the noise level, it stays at 85. sound meter was at 110 dB.

[Julian Bunn]

Are there any audio system settings on the phone that you can play with? This shouldn't be necessary, but now I'm wondering what Samsung have done!

Julian

On Thu, Jul 23, 2020 at 2:11 AM <tonko....@gmail.com> wrote:

with the front mic is the same. it can only measure up to 85 dB. althogh i incresed the noise level, it stays at 85. sound meter was at 110 dB.

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/4d4700c2-3a21-4847-9d97-1bab813f0fd8o%40googlegroups.com.

[tonko....@gmail.com]

there is no setting on a41 to switch compression off

[Julian Bunn]

Is the Dolby Atmos setting on?

From the manual: "On the Settings screen, tap Sounds and vibration → Sound quality and effects → Dolby Atmos, tap the switch to activate it, and then select a mode."

On Thu, Jul 23, 2020 at 8:57 PM <tonko....@gmail.com> wrote:

there is no setting on a41 to switch compression off

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/19fe7438-f59c-44ed-86bf-c845085ae4c6o%40googlegroups.com.

[tonko....@gmail.com]

it was off. i tried now with on. it is the same result

[Julian Bunn]

I'm sorry to hear that, and am out of ideas. It seems that the firmware in the A41 is preventing AudioTool from getting raw microphone samples without filtering or AGC.

Julian

On Thu, Jul 23, 2020 at 11:15 PM <tonko....@gmail.com> wrote:

it was off. i tried now with on. it is the same result

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/59845210-54f2-4623-b81b-a5145ef82de5o%40googlegroups.com.

[Jean-Baptiste ROMANET]

Hi,same problem since years one a Oneplus 3T smartphone.
The only workaround is USB audio interface.

I didn't know there was a specific audio flag prevent AGC. I will try to find a solution on this side as my phone is rooted

Jean-Baptiste

[tonko p]

today i connected dayton mic to samsung a41 through usb-c audio interface. no compression occurs this way. i made test with white noise and flat evaluation. it doesn’t work under approx. 65 db, but it works perfectly from 65 to 105 db (this was max noise level i could generate).

[tonko]

Hello,

if it is possible to avoid compression on Samsung phone microphone, please let me know how.

Regards

Tonko

--
You received this message because you are subscribed to a topic in the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/audiotool-discussion-group/h1a1Ndry4po/unsubscribe.
To unsubscribe from this group and all its topics, send an email to audiotool-discussio...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/0e19d22e-07ef-479c-bf92-bf1ffc751380n%40googlegroups.com.

[Julian Bunn]

AudioTool uses an audio data stream that is supposed to be a clean, unfiltered, uncompressed, unmodified set of samples from the microphone. There should be no compression.

Julian

You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CA%2B%3DM%3DWxV_XPiw6zssS4PMVbaqNe-U4gf__QHH8L6REmiJXMW%3Dw%40mail.gmail.com.

[tonko]

In that case, I must be doing something wrong. 

Phone microphones on my both Samsung phones, A41 and S24, measure white noise level 50 dB - the same as sound meter and Samsung Tab S2 with Dayton iMM-6.

But phone microphones A41 and S24 measure about 90 dB, wenn sound meter and Tab S2 with Dayton iMM-6 measure above 100 dB.

Tonko

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAAxkZQsis2zjsoR0840dCH4ykvSwJAdGijpNmUtzFJSSqH_-LQ%40mail.gmail.com.

[tonko]

... and aditional information. With Decibel X app, all 4 devices measure the same level at 50 db as well at 100 dB.

[Julian Bunn]

It sounds like a) Samsung have implemented some sort of compression or filter on their implementation of the "VOICE_RECOGNITION" mic audio stream on the A41 and S24 devices, and b) Decibel X app is using some other audio stream (not VOICE_RECOGNITION) for their processing, so they are always getting filtered data.

What does a Pink Noise spectrum look like as measured on the A41 and S24?

I see that there is a new mic category in the SDK called UNPROCESSED: UNPROCESSED

Microphone audio source tuned for unprocessed (raw) sound if available, behaves like DEFAULT otherwise.

So it may be that AudioTool should switch to using that now. I can make a test APK for you to try to see if it makes a difference, if you'd like?

Julian

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CA%2B%3DM%3DWx3gJTkTHcMk2AeKzEdSM2oofxYGhY%3Dv-5iGzoAJmbSNQ%40mail.gmail.com.

[tonko]

Hi Julian,

today I tried to calibrate Decibel x on my both Samsung phones. Works fine on A41 but doesn't work on S24 (mic is compressed). See attachment.

Tonko

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAAxkZQsXAMUbqjrbYORn2H33Gx0eYzEmhPoeNg-ROvh94kc7bQ%40mail.gmail.com.

[Julian Bunn]

I'm afraid I can't comment on what that app is doing - it's not AudioTool.

Julian

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CA%2B%3DM%3DWyOH%2B-6s6WXOR5troTuvXmp8kbguxrB%2BHREcd36xzN%3DFQ%40mail.gmail.com.

[Jean-Baptiste ROMANET]

😉

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAAxkZQtF7xD280xiEs_h22htVw%3DShezJGastZ1-%3D92FVBm42Bw%40mail.gmail.com.

[tonko]

My mail was relatid to my previos mails. I thought this info could be interested for you because AudioTool also doesn't work with S24 mic.

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAAxkZQtF7xD280xiEs_h22htVw%3DShezJGastZ1-%3D92FVBm42Bw%40mail.gmail.com.

[Julian Bunn]

Yes, I understand. In my previous reply I offered to send you a special AudioTool APK that might fix the problem on that phone. Let me know if you are interested.

Julian

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CA%2B%3DM%3DWyS%3DEVHFWn4fN1Sxv0TR47h6AJYez7u7fGPNMvk-0K3YQ%40mail.gmail.com.

[tonko]

Yes, please send me this version.

Tonko

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/CAAxkZQu_BYX9aEW437ScKdC7sB-Zt8vtQJTi0ROiLgMM-npfXg%40mail.gmail.com.

[pa sk]

Hello Julian,

Hello Tonko,

About the "unprocessed" stream: such named audio sources have been present for good couple of years in two apps I use - Hi-Res Audio Recorder v. 0.10.78 from 2019 and Spectroid v. 1.1.2 from 2022.

When recording audio using this source instead of the other available, including the "voice recognition", indeed makes a difference.

When analysing audio signals (Spectroid) the difference is obvious in spectra and huge in the dynamic behaviour but only when using internal mikes - external input behaves the same under both settings.

All the above on POCO F1, Android 10

I attach some files with spectra of few measurements and some texts - sorry for this spam-like action.

Btw, my experience with several different phones show that samsungs are the most unpredictable and the most intrusive in processing (I checked A51, S8, Motorola Edge 7, Honor 8s, Huawei P10, POCO F1 and some others but very briefly).

First three graphs "show" the same sound (fridge in the kitchen :) ), following two the - same kitchen, fridge off ;).

Regards,

Paweł S.

[Julian Bunn]

Thanks for sharing this very interesting investigation!

So, it looks like if using an external Mic, then the response is the same whether or not "VOICE_RECOGNITION" or "UNPROCESSED" is used, whereas if using the internal Mic, then there is a difference, and the UNPROCESSED is better?

I may then change AudioTool's default to use UNPROCESSED instead of VOICE_RECOGNITION.

Julian

To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/6654073b-5048-499b-830b-fb07ea6b9dc2n%40googlegroups.com.

[pa sk]

You know better than me that android farm/ecosystem is as diverse as can be and the best assumption is to treat each phone (HW + SW) as an individual beast.

That said, yes, it looks like the "unprocessed" AKA "raw" source is better than any other from metrological POV. (and here a huge respect for even thinking such a view may exist in case of phones internal MEMS mikes :O

In addition, in case of POCO F1 on Android 10, the differences between "unprocessed" and "voice recognition" disappear when the external mike is connected to audio jack.

The main improvement (all about internal mikes) is the elimination of what the guy from Hi-Res Audio Recorder called auto-gain and what I would rather describe as the frequency dependent dynamic gain.

The bandwidth is also less cut in the lower freqs from 200 Hz downwards (on Motorola Edge+ and Honor 8s the lower end was cut in either case, although to the lesser extent and differently than in POCO F1).

But, be frank, work with audio without the mike of defined characteristics and the linear signal path (amplitude, frequency, time) is a work in vain. (yep, the subject for a longer discussion, I know)

Regards,

Paweł S.

[pa sk]

Hello Julian,

As a kind of supplement to what I wrote:

Hope it is fairly self-explanatory.

Columns and chart contain calibration coefficients for two different phones.

Pictures show spectra of a quiet room background noise caught with iMM6 mike through different audio streams.

Conclusions are two: 1. "unprocessed" stream is closer to reality, 2. no hope for any unification ...

(calibration files are the results of the measurements I did using calibrated sound pressure level meter)

Regards again,

Paweł S.