Audiotool + Dayton IMM-6 on Galaxy S3 high pass filter?

[e-m...@jurriespoelstra.nl]

I use JJBunn's Audiotool on a Samsung S3 with a Dayton IMM-6 external mic and managed to load the calibration file from Dayton. I then tried it side by side with a calibrated iPhone with Studio Six Digital Audiotools + MicW i436 on a outdoor live show.

On Android with JJB AT it showed the low end being rolled off at around 200Hz and not showing much action. The iPhone clearly showed peaks around 50Hz or 105Hz where on the S3 there wasn't much happening. Also it looked as SPL metering would max out at around 102dB(C), where the show measured 107dB(C) at times. Furthermore I noticed that the SPL reading would colour red when high. Maybe there is overloading at the input or limiting going on?

On the Studio Six website I read that certain iPhones have high pass filtering at the input to clean up talking voices in a loud environment. And that they manage to turn it off with their software. Might this be the case with Samsung S3 as well?

Does anyone know of low-end roll off on the Galaxy S3?

Does JJBunn's audiotool provide measuring from 20Hz and up on the S3?

Might there be a maximum input of 100dB on the Galaxy S3 that's causing problems?

Thanks for your time.

J

[Julian Bunn]

Hi,

A couple of comments/observations:

1) Even with a calibrated mic such as the MicW or iMM6, one is still limited by the audio input circuitry in thee phone. Some phones are better than others.

2) On iOS the programmer can specify a "measurement mic" as input, which is supposed to defeat any AGC or filtering - this is what iAudioTool (the iPhone/iPad version of AudioTool) does, and I suspect what Studio Six are referring to.

3) On Android, there is an equivalent setting, which AudioTool uses.

4) 16bit samples limit the dynamic range to 96dB on the input. 107dB is going to produce clipping unless the input is attenuated somehow. On Android and iOS, 16 bit samples are used by (i)AudioTool. At some point, 24bit samples may be available in the APIs, and that would significantly increase the dynamic range.

It's possible that the S3 is not a good choice of phone for demanding audio measurement - I do hear reports from some users of significant roll-off in the low end, but I have a feeling that this is model/Android OS version related. 

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.
For more options, visit https://groups.google.com/groups/opt_out.

[e-m...@jurriespoelstra.nl]

Hi Julian. Thank you for your answers.

I sent an e-mail to Samsung asking if they have a low freq roll-off limitation on the i9300. I hope they will reply soon. I found out that the low endy roll-off limitation is not present in the chip that's being used, so could it be filtering in the Samsung OS shell that can't be turned off or something? Is that possible? Android does give you access to unadulterated audio input stream, doesn't it? It might also be just occuring at high SPL. I'll measure some more soon. Furthermore I read on the Studio Six Digital website that the iPhone SPL can surely reach 107dB as I've seen it do. But how is that possible with 16-bit samples?

J.

Op dinsdag 20 augustus 2013 03:57:10 UTC+2 schreef Julian:

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussion-group+unsub...@googlegroups.com.

[Julian Bunn]

Yes, the Android SDK allows access to an unfiltered audio stream, which is what AudioTool uses.

To get to 107dB is quite easy ... just apply a suitable calibration constant! The default internal value in AudioTool is ~20dB (there's a small correction for the FFT windowing function), which effectively means that SPLs in the range 20dB to 116dB (i.e. 96+20) are shown. The problem comes when there is clipping on the mic input circuitry, of course - in that case the input samples are at 96dB relative to their zero value, and depending on what the calibration is, this could read 96dB, 116dB or whatever.

So, the important point is that because these devices use 16bit ADCs, the *dynamic range* is fixed at 96dB, but the maximum dB shown can be arbitrary.

Julian

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

[e-m...@jurriespoelstra.nl]

Hi Julian,

I managed to get SPL readings above 100dB. I haven't calibrated it with a pro device, so the reading is probably not right.

In the attached screenshots you can see the low end roll off. The measurement was taken in a small club where there was a huge amount of low end. 

Any idea on how to bypass the lowend rolloff on the Galaxy S3?

Kind Regards

Op donderdag 22 augustus 2013 17:51:06 UTC+2 schreef Julian:

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussion-group+unsubsc...@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

[Julian Bunn]

Hi,

Thanks for this, and the great screenshot. It is evident that there is a significant roll off on the low end. I have just had an opportunity to test AudioTool on an S3 myself, and didn't see this as anywhere near as pronounced. That suggests it is dependent on the firmware or some system setting. 

What Android OS version are you running? I will compare with the S3 I have access to here.

Julian

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

<Screenshot_2013-08-29-00-10-53.png>

[e-m...@jurriespoelstra.nl]

I am running Android 4.1.2, kernel version 3.0.31 - 1328525, buildnumber JZO54K.I9300XXEMG6. The screenshot was with the Dayton. I'm off to a small outdoor show now and will take another screenshot there to see if it also occurs with reasonable volumes.

J

Op vrijdag 30 augustus 2013 17:06:24 UTC+2 schreef Julian:

<Screenshot_2013-08-29-00-10-53.png>

[Julian Bunn]

The S3 I have here is slightly earlier: 3.0.31 - 1314436 and ...XXEMG4. This phone does not show such a pronounced falloff as yours, but it is definitely dropping steadily below ~125Hz.

I think that Samsung are applying some sort of high pass filter. 

Julian 

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.

[e-m...@jurriespoelstra.nl]

Around 90dB it looks more like the rolloff starts around 125Hz, just like yours. The high SPL might have something to do with the difference in readings.

Unfortunately Samsung never replied to my Q's on HPfilter in the S3. Do you think there might be a chance you and/or other developers will find a way to bypass their filtering?

J.

Op vrijdag 30 augustus 2013 21:10:32 UTC+2 schreef Julian:

<Screenshot_2013-08-29-00-10-53.png>

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

[Julian Bunn]

A sub-optimal way of dealing with the filter is to apply a correction using 1/3 octave calibration - just specify calibration offsets for the sub-125Hz bars that restore the response to flat with Pink Noise fed in to the mic input. The problem with this is that at the very low frequencies it will just be boosting electronic noise, probably.

If AudioTool was only targeting the S3 device, then it might be possible to defeat their filter by directly controlling the hardware via the NDK - but I really don't want to go there as it makes software maintenance a nightmare :-)

Julian

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.