11 Jun
2010
11 Jun
'10
12:28 p.m.
Hello,
I have developed an audio recording application for Android, and a user is
reporting a weird noise issue. There are near-zero chances that this comes from
my code, and I'm looking for the possible cause of these noises.
Here is the sample, it's a musical flute and violin recording:
http://sound.samalyse.org/tapemachine/pulsations.wav
Listen carefully and you will here small pulsations.
I've also asked him to record a 440Hz sine. This one isn't distorted:
http://sound.samalyse.org/tapemachine/record440hz.wav
I suspect that this is a hardware issue, but I'm no expert so I can't really
tell. The phone is an HTC Incredible, which unless I'm wrong, means that the
sound components are embedded into a Qualcomm Snapdragon QSD8650 chip.
One other is that, although Android is a linux-based system, and use alsa
underneath, there's a terrible high-level Java API, which is full of various
problems which I have worked around to obtain that kind of quality. Thus, I
can't really tell what happens in there.
Have you already heard such things, and know what the cause, whether hardware or
software, is likely to be?
--
Olivier
11 Jun
11 Jun
2:10 p.m.
Olivier Guilyardi wrote:
Here is the sample, it's a musical flute and
violin recording:
http://sound.samalyse.org/tapemachine/pulsations.wav
Listen carefully and you will here small pulsations.
Here I didn't need to listen carefully, it's unmistakable. I know this
kind of noise when doing recordings with anlog consumer cassette
recorders and cheap microphones and cables.
Listen carefully and you will here this or similar pulsations for the
sine wave too, perhaps just some background noise, when the sine was
recorded.
What happens for recordings without microphones?
Ralf
2:44 p.m.
On 06/11/2010 02:10 PM, Ralf Mardorf wrote:
Olivier Guilyardi wrote:
Yes, you're right, the noises are obvious.
Here is the sample, it's a musical flute and
violin recording:
http://sound.samalyse.org/tapemachine/pulsations.wav
Listen carefully and you will here small pulsations.
Here I didn't need to listen carefully, it's unmistakable. I know this
kind of noise when doing recordings with anlog consumer cassette
recorders and cheap microphones and cables. Listen carefully and you will here this or similar
pulsations for the
sine wave too, perhaps just some background noise, when the sine was
recorded.
I think these are background noises. It has been recorded by putting the phone
near the computer speaker. I'd swear that this is murmuring in the background.
What happens for recordings without microphones?
There is no line-in, only a TRRS jack for both input and output. So I don't see
how to record without a mic. The user has tried to plug a headset with an
integrated microphone, and he said that the noises were still here.
He also reported the following: "if I rig up a TRRS 3.5 mm jack to XLR adapter
and use a "real" microphone the recording seems to be clean (though with other
issues, perhaps ground loops or impedance mismatch or something)".
I've seen similar problems when plugging an external microphone. That produced a
huge constant noise. I doubt that the user can really determines if the
pulsations are gone in this situation.
On the software side, couldn't a sample rate conversion produce that kind of
pulsations? I'm not doing any, but I haven't checked Android sources about this.
--
Olivier
3:13 p.m.
Olivier Guilyardi wrote:
He also reported the following: "if I rig up a
TRRS 3.5 mm jack to XLR adapter
and use a "real" microphone the recording seems to be clean (though with other
issues, perhaps ground loops or impedance mismatch or something)".
*!*
On the software side, couldn't a sample rate
conversion produce that kind of
pulsations? I'm not doing any, but I haven't checked Android sources about this.
I don't have knowledge about this and at the moment I'm a little bit
pissed of all the issues we could get, when using the computer ;).
I tend a little bit to prefer analog equipment at the moment, but your
noise issue could be an anlog issue :D.
Regarding to this or similar strange pulsations, I heard this for
recordings done with cheap consumer microphones and cables using analog
equipment.
No doubt, there are all kinds of issues possible, regarding to the
digital domain, but this noise or a similar noise is something I *very
often* heard for recordings done with (very) cheap consumer equipment.
:)
Ralf
3:53 p.m.
On 06/11/2010 03:12 PM, Ralf Mardorf wrote:
Olivier Guilyardi wrote:
Old school mood I see ;)
He also reported the following: "if I rig up
a TRRS 3.5 mm jack to XLR
adapter
and use a "real" microphone the recording seems to be clean (though
with other
issues, perhaps ground loops or impedance mismatch or something)".
*!*
On the software side, couldn't a sample rate
conversion produce that
kind of
pulsations? I'm not doing any, but I haven't checked Android sources
about this.
I don't have knowledge about this and at the moment I'm a little bit
pissed of all the issues we could get, when using the computer ;).
I tend a little bit to prefer analog equipment at the moment, but your
noise issue could be an anlog issue :D. Regarding to this or similar strange pulsations, I
heard this for
recordings done with cheap consumer microphones and cables using analog
equipment.
No doubt, there are all kinds of issues possible, regarding to the
digital domain, but this noise or a similar noise is something I *very
often* heard for recordings done with (very) cheap consumer equipment.
Okay, that's precious informations. I wasn't sure this could come from the mic,
but you're making it clear that it's much probable.
By the way, this is all about low-end (or very miniaturized) stuff, but also
real world: a lot of musicians find it handy to have a recorder in their phone.
Of course that's more about recording rehearsals than concerts.
--
Olivier
7:43 p.m.
Olivier Guilyardi wrote:
Okay, that's precious informations. I wasn't
sure this could come from the mic,
but you're making it clear that it's much probable.
At least similar (maybe the same) noise isn't an unexpected surprise,
when using combinations of low-end analog consumer equipment. You anyway
should have a look at your source code. I experienced digital noise
similar to this more aggressive, excepted of some digital transient
noises, IIRC when using Rubber Band to pitch something in real-time,
while the length, timing is kept.
Ralf
--
Does anybody know were the Roland VR 760 currently is sold in Germany?
12 Jun
12 Jun
12:10 a.m.
Olivier,
hard to tell from such a small fragment if it's regular or random.
it maybe irq swapping, or overload compression built into the card or...
what card is it? describe the signal chain.
try recording at 1/2 level and see if the problem decreases/stops. it's
more prominent when the violin enters
g.
13 Jun
13 Jun
12:30 p.m.
On 06/12/2010 12:10 AM, Geoff Beasley wrote:
Olivier,
hard to tell from such a small fragment if it's regular or random.
it maybe irq swapping, or overload compression built into the card or...
what card is it? describe the signal chain.
I don't know which sound card/chip it is. It's running on a phone, an HTC
Incredible, which has the following specs:
http://www.htc.com/us/products/droid-incredible-verizon#tech-specs
The main chip, which embeds the 1Ghz CPU and a lot of other things, is a
Snapdragon QSD8650:
http://pdadb.net/index.php?m=cpu&id=a8650&c=qualcomm_snapdragon_qsd…
This is a bit of a do-it-all chip, so I suppose it might embed sound components.
But there's nothing sure about this. Maybe that dmesg could help, but it's a bit
complicated to do this with a remote user.
The signal has being recorded using the internal mic, but the user reported that
the same problem occurs when using a headset with integrated mic.
try recording at 1/2 level and see if the problem
decreases/stops. it's
more prominent when the violin enters
It's a report from a remote user. I'm already waiting for some more feedback
from him about testing with an external dynamic mic.
Also, about level, Android doesn't give me any access to the hardware mixer. So
the only way to decrease level would be to record from a greater distance.
From your other mail:
another thought is resampling if it's a 48k card running @ 44.1 it may be resampling
on the fly...
I'm reading audio PCM data through the Android Java API, which is a bit abstract
but doesn't convert the native ALSA samplerate AFAICS.
Right now, I highly suspect that the HW performs some automatic gain control or
so, suitable for phone communications, but altering the signal in this case.
--
Olivier
5431
days inactive
5433
days old
linux-audio-dev@lists.linuxaudio.org
7 comments
3 participants
participants (3)
-
Geoff Beasley -
Olivier Guilyardi -
Ralf Mardorf