OT: Why do people shout into mobile phones?

The same stupid people are here in Sweden.

If the intent is for clarity then woefully wrong - overloads the system/compression/codec or whatever is in the transmission system so clipped/semi-garbled voice at the receiving end

People are used to hearing themselves in the earpiece when talking. In telco terminology, it's called "sidetone".

The problem is that there's about a 250 msec audio delay inherent in digital cellular phone encoding that will drive the user nuts if there's even the slightest leakage between the sidetone and the microphone or through the system. Analog cellular did not have the delay. At best, there will be an echo. More typically, there is massive confusion when the user hears themselves somewhat delayed. In addition, the level of side tone necessary to convince the typical user that they're getting through, is insufficient to prevent leakage back into the microphone, especially with a BlueGoof headset. If the handset manufacturer or cellular vendor added sidetone, you would instead be complaining about either feedback squeal if injected locally in the handset or echo if injected at the switch.

Another contributory phenomenon is cellular garble causes people to think they're not getting through, so people normally talk louder to compensate. The more garble, the more louder they yell.

One interesting observation is that few people yell into a half-duplex phone, or phone that requires a PTT (push to talk) to be heard. That's because there's no expectation of ever hearing oneself in the earpiece.

I've had a cell/mobile phone of sorts since before IMTS, so I'm quite familiar with the problem. I've taught myself not to yell and have convinced a few others to make the attempt. It does take some practice, but is easy enough. The problem is convincing other people that it's a problem. I have a digital sound level meter and sometimes drag it into a restaurant for some entertainment. Most people are rather surprised when I mention that they're yelling anywhere from 3 to 6dB louder into the phone than their normal conversation (while talking to me). Although I expected considerable resistance from my waving sound level meters at people, most are generally curious and interested in the effect.

Also, for truly irritating cell phones, the Nextel IDEN phones with the *VERY* loud speaker, is the worst. While suitable for noisy construction sites, having one of those go off in a restaurant is the epitome of obnoxious behavior.

There are some really interesting comments and observations in this article (4 pages):

More. (NPR audio 5 min):

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

Won't happen. In the handset, between the microphone and the A/D converter (CODEC) is an AGC (automagic gain control) circuit with considerable dynamic range. No matter how loud one yells into the microphone, the AGC makes sure that the audio input to the A/D converter never sees more than the peak to peak level necessary to use all the digital bits available. It is possible to overload the AGC using an external audio device (i.e. headset), but those are designed to also deliver a fixed maximum output.

Here's a simple test you can do yourself. Call your answering machine from your cell phone. Do NOT listen to your answering machine while doing this test. Speak a test phrase, starting with a very low voice level and with the phone perhaps 20 cm away. Move closer in stages. Talk louder in stages, until you're eating the microphone and screamiing into the phone. Play back the answering machine message and compare results. If you're ambitious, attach an audio level meter or scope to the answering machine speaker. I think you'll find that except for the first low level test, they'll all be almost identical in volume with perhaps a 3dB spread. Clarity and distortion might be a problem at the screaming end of the test, but I think you'll find that there's no garble because no bits were lost during the test transmission.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

[Jeff L.]

Wouldn't it be simple to amplify the speaker's voice, locally, and mix that into the receiver circuit for immediate sidetone? No echo, no delay, no yelling.

Yes, battery life would suffer a bit, but everyone would benefit. :-)

One of those things that phone manufacturers would resist, and only legislation would make happen.

"DaveC"

** You ACTUALLY deleted the answer to your question.

" In addition, the level of side tone necessary to convince the typical user that they're getting through, is insufficient to prevent leakage back into the microphone, especially with a BlueGoof headset. If the handset manufacturer or cellular vendor added sidetone, you would instead be complaining about either feedback squeal if injected locally in the handset or echo if injected at the switch. "

This was immediately after the stuff you quoted.

Wot a moron.

.... Phil

Yes, it would be "simple" but not very functional. The problem is that you would also need to limit the microphone AGC range in order to prevent the sidetone circuit from becoming an acoustic feedback generator.

In addition, the background noise cancellation circuit will also be affected. Some of the better cell phones have two microphones. One for you to yell into. The other to pickup background noise. These are combined 180 degrees otto phase, resulting in the cancellation of background noise. Any sidetone that leaks into either microphone is not going to be balanced with the other microphone, resulting in feedback.

In addition, the sidetone will need to be disabled when operating as a speakerphone. That's no problem because everyone yells into a speakerphone.

However, there is a way to make local sidetone sorta work. A slight frequency shift (Bode shift) between the microphone audio and the sidetone audio will prevent feedback. This is what's done on POTS phones and some VoIP system. You can't hear a 5-10Hz frequency shift, but it will prevent feedback. (Incidentally, when presented with a dialup modem tone, the telco echo canceller is disabled. You may not be able to hear a 5Hz shift, but it will drive the modem nuts).

However, that only solves the feedback problem. The background noise canceller and speaker phone operation will still need to be dealt with, or tolerated.

So, why isn't this being done? In order to make it work, one would need a separate DSP circuit for the sidetone, which is both expensive possibly messy. Because it's NOT a complete solution, it hasn't been used. However, it certainly has been tried.

A little audio has nothing to do with battery life. The biggest battery hog in a cell phone is the transmitter. 2nd biggest is the display backlighting. Unless it's a Nextel loud blaster, the audio circuitry is well down the list.

Incidentally, if you need a demonstration of how much the transmitter sucks power, put your cell phone in a shielded box, such as a microwave oven. Presumably, the phone cannot hear the cell site inside the box. However, it will almost continuously announce it's presence and listen for a response from the cell site. Average runtime from fully charged to totally dead seems to be about 4-5 hrs when I tried it last.

I think I'm going to lose my dinner. Do you really believe that a horde of elected officials are even capable of understanding technical issues? Government is a problem, not a solution.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

"Jeff Liebermann"

** No way does a 5-10 Hz frequency shift *prevent* acoustic feedback !!

That is a very dopey myth.

The effect of a few Hz shift is to modify the familiar, single frequency howl into a warbling tone.

The exception is where there is a lot of reverberation in the path from speaker to mic, causing acoustic gain at multiple specific frequencies. Then a few Hz shift can produce a useful increase in the threshold before any disturbance (like ringing) occurs.

..... Phil

Frequency shifting has been used in public address systems to help control feedback since the 1960's.

The local auditorium has a nifty Bode shifter that works well enough with some limitations. See below.

Maybe. In an uncontrolled environment, it's possible for there to be positive feedback at specific frequencies and delays. For music, the maximum frequency shift is 3Hz or less as the ear is very sensitive to harmonics, which are not quite right with a greater than 3Hz shift. Piano harmonic partials sound particularly awful. 3Hz is also insufficient to prevent feedback under all environmental circumstances. However, for voice quality (PA and cellular), 5-10Hz works quite well. The current fashion is to continuously vary the frequency shift. If there are any frequencies where there is positive feedback, the length of time that the system spends at that frequency is insufficient for feedback or howl to develop.

Yep. That's the same thing that I just mumbled. My point is that it can be done with voice and cellular. Music is more problematic.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

"Jeff Liebermann" "Phil Allison"

** Has nothing to do with your false assertion.

Plus I specifically MENTIONED that use in my post.

** No maybes about it - it IS what happens, asshole

I have designed a high quality shifter for PA system use - details were published in Electronics Australia magazine in August of 1997. The maximum benefit is only 6 to 8 dB in the gain before feedback threshold when the PA is used in a very reverberant room, like a church.

No benefit exist where *direct path* feedback is the culprit.

** There NO similarity between the two apps.

Never noticed how PA systems are often in large buildings with long reverberation times ???

** No asshole.

YOUR words came along in parody of MINE !!

** Frequency shifting is of help with VOICE in a PA system - ie where the " PUBLIC " are being " ADDRESSED" by a speaker. Hence the term.

No way does a 5-10 Hz frequency shift *prevent* acoustic feedback in general.

Certainly not when the mic and loudspeaker are only inches apart.

..... Phil

For the record, my alleged "false assertion" is that a Bode frequency shifter can be used to reduce or eliminate acoustic feedback in a cell phone.

Ok, I'll give you credit for getting that part right.

My hemorrhoids are none of your business.

I tried to find some details on your design, but only found a thread where you mentioned it in rec.audio.pro.

I couldn't tell what you used to implement your design. Analog with 2 multipliers and two oscillators 5Hz apart?

I beg to differ. At the local auditorium, it's common for the speaker to waltz in front of the loudspeakers with microphone in hand. No feedback at all. However, when I listen carefully scope, I can hear reflections from the opposite wall. The only screwup seem to be when the microphone is very directional and pointed at the opposite wall. That sounds like a dying Klingon warship.

Agreed. PA has to deal with the room acoustics. A cell phone doesn't have enough speaker acoustic power for the room to make any difference. In any case, the users head is going to block any room reflections.

I don't do commercial audio, so the answer is no, I have never noticed. However, in my checkered past, I helped build and debug 2 recording studios. My job was getting the hum out of the system but I also helped with the room acoustics. We didn't have Bode shifters at the time, so I was not able to try one in a studio.

I didn't read your concluding paragraph until after I scribbled mine. I added a little detail explaining how it worked, something lacking in your explanation. Rather than just erase what I wrote and agree with you, I gave you credit and left it as is. Try really hard not to assume that every line I scribble disagrees with everything you write.

Well, I'll admit that I haven't built a DSP audio processor with a 5Hz shift suitable for testing on a cell phone. I might be totally wrong, but you haven't bothered to supply a reason why I'm wrong about it working in a cell phone.

If you throw in room acoustics, you're right. If you don't have to deal with room acoustics, because you head is blocking the earphone, it's not a problem and Bode shifting will help with feedback.

Actually, it works better when there's a direct path. As you indicated, it's the reflections that cause problems.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558