Jump to content


Photo

Analysis of the HTC-Soundbug (tsss-noise)

- - - - -

  • Please log in to reply
68 replies to this topic

#1
bazzo

bazzo

    Regular

  • Members
  • PipPip
  • 96 posts
  • Devices:SDAmusic
Analysis of the audio-bug on HTC Smartphones
...also known as "tsssss-noise"

Summary of the test results

Test 1: Which relevance has the audioformat/compression?
A: It is not relevant.

Test 2: Are there regularities in the apperance of the bug?
A: Yes, the bug appeares all 16,680s and lasts 110ms.

Test 3: Which influence does the frequency of the input signal have on the level of the noise?
A: The level of the noise increases with increasing input frequency logarithmically proportionally.

Test 4: Which influence does the Freq. of the input signal have on the spectrum of the noise?
A: There are characteristic frequencies developed, additionally a whitenoise is superposed.

Test 5: Which influence does the level of the input signal have on the noise?
A: The level of the noise increases equivalent with the level of the input signal.

Test 6: Which influence does the betaPlayers Equalizer have on the noise?
A: It does not have an influence.

Test 7: Which influence do other settings of the test-smartphone have?
A: I found no relevant settings for the bug.

Test 8: Which influence does the quantization of the audio data have?
A: It does'nt have an influence.

Test 9: Does the noise also a occur when playing back via internal loudspeaker?
A: Yes.

The detailed test
The aim of this analysis is the more accurate identification of the, from many users discovered, "TSSSSS" noise with the mp3-playback on HTC Smartphones. This noise makes the music unenjoyable. Affectet devices are only usable als Audioplayer in a streetcar.

Test-setup:
The test equipment was a SDAmusic (aka "HTCamadeus"), it is attached to the line-in of the computers soundboard. As test software I used a 30day-trial of Adobe Audition. The test-smartphone was always in flightmode to exclude interferences by the GSM module.

Execution:
All tests were accomplished as follows. First, an audio test-file was generated with the AdobeAudition noise generator. This file was then transferred in each case to the test-smartphone and played back with the betaPlayer. Its Equalizer was always switched off, if explicitly not mentioned, and the volume of the test-smartphone was adjusted to maximum.
The audiosignal was recorded and analyzed, over the soundboard, with Adobe Audition. The input of the soundboard was adjusted, so that the recorded signal has nearly the same level as the generated input file. Further details on execution are to be found with the appropriate tests.

In first pre-tests, it turned out that a very high sinetone is well suitable to provoke the tssss-noise.


Test 1
Which relevance has the audioformat/compression?


Testtone: 16000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo

This testtone was stored in different files with different compressions.

Testfile formats:
mp3(lame3.90.2 --alt -preset standard)
wma (128kbps CBR)
wav (PCM uncompressed)

Result:
The noise occurs, with all testfiles, with same level. Since it is audible with the uncompressed wav-file, I assume the compression does not have influence on the error. A software fault of the mp3-decoders is to be excluded therefore.

Test 2
Are there regularities in the appearance of the bug?


Testtone: 16000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo

Testfile formats: wav (PCM uncompressed)

I used "Auditions" notch-filter to eliminate the input sinetone in the recording. Otherwise no further manipulations were accomplished.
(for the screenshot I have deleted everything before the first noise, therefor the beginning of the bug-noise starts at 0ms on timeline.)
* "notch filters" - eliminates extremely narrow-band a certain frequency

1.The length of the noise

noise no. 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
length(ms) 114|110|109|112|111|116|110|112|114|110|

Based on 10 mesurements, there is a duration of approx. 110ms. It is remarkable that the noise appeares 55ms on the right channel and only thereafter 55ms on the left channel (see fig.1)

Posted Image
fig.:1 Waveform of the disturbance after notch filtering of 16kHz

2. Spacing between the noises

noise no. 1-2 | 2-3 | 3-4 | 4-5 | 5-6 | 6-7 | 7-8 | 8-9 | 9-10 |
intervall(s)|16,679|16,681|16,679|16,680|16,680|16,680|16,681|16,680|16,681|

I've mesured the intervall between the beginning of a noise and the beginning of the respective next noise. The interval is on an average of 16,680s.

Posted Image
fig.:2 intervall between 4. and 5. noise (consider time code in the field ("Sel.";"Begin") = position of the marker)

Result:
The noise appears all 16,68s, when continuous playing the test-file, and lasts approximately 110ms(55ms left, then 55ms right).
During the interrupted playback, thus with intermediate pausing of the playback, the intervall between the noises changed. There are no more regularities recognizable. Within continuous played passages, the intervall between the noises was however again 16,68s. The length, of 2x55ms, of the noise did not change.


Test 3
Which influence does the frequency of the input signal have on the level of the noise?


Test tones:
16000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
8000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
4000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
2000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
1000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
500Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
250Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
50Hz-22kHz whitenoise (level -3dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)

In these resulting recordings, I eliminated the sinetone by using the notch-filter with the particulare frequency. The maximum levels were mesured in the waveform, in each case 3 noises and averaged afterwards.

input freq.(Hz) 16000 8000 4000 2000 1000 500 250
noiselevel (dB) -2, -6, -12,-16,-24,-27,-37

Posted Image
fig.: 3 level of the noise over the logarithmical inputfrequency

translations:
"Eingangsfrequenz"= input frequency
"max. Pegel der Störung"= maximum level of the noise


Result:
It becomes clear that the level of the noise rises linear with the logarithm of the input frequency. It is particularly remarkable that the interference exceeds the level of the input signal (-3dB) starting from a frequency of 16kHz. The peaks of the noise are to be already recognisable before notch-filtering the sinetone.
When playing the file with the whitenoise, the disturbance wasn't measurable. The "noise filter" left no disturbance. If there was a noise at all, it went down in the backgroundhiss.


Test 4
Which influence does the frequency of the input signal have on the spectrum of the disturbance?

Test tones:
16000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
8000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
4000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
2000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
1000Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
500Hz sinetone (level -3dB) - 44,1kHz/16b/stereo
250Hz sinetone (level -3dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)


Posted Image
Fig.: 4 spectrum of the noise at 250Hz

Posted Image
Fig.: 5 spectrum of the noise at 500Hz

Posted Image
Fig.: 6 spectrum of the noise at 1000Hz

Posted Image
Fig.: 7 spectrum of the noise at 2000Hz

Posted Image
Fig.: 8 spectrum of the noise at 4000Hz

Posted Image
Fig.: 9 spectrum of the noise at 8000Hz

Posted Image
Fig.: 10 spectrum of the noise at 16000Hz

The figures(4-10) show the spectrum of the noise (RED) compared, with that spectrum of the continuous backgroundhiss(BLUE) which is superposed all recordings. I assume that the peaks recognisable in the noise and in the hiss, result from an overdrive of the smartphones amp or the soundboards line-in.
This clipping simply cuts the peaks off of the sine function and produces higher frequency portions (furrier-progression-components of higher order) which are not effected by the notch-filter. These peaks are normal in this case and have nothing to do with the bug.
It is remarkable that into far divide the spectrum of the noise is congruent. It seems that the bug simply amplifies the hiss, but thats not the truth. The hiss is not added until the recording. Therefore is to be assumed, that a part of the noise is a "whitenoise", which increases accordingly with increasing frequency (test 2).
Another much strange part of the noise is characterised by sinetone newly developed. Considering e.g. the pair of peaks in (fig.: 4) around 9500Hz, some similar peaks are also with 6000Hz, 15500Hz and with 19000Hz clearly recognizable. These peaks seem to be pushed apart with increasing frequency of the input signal (see fig.:5-10) around exactly the amount of the frequency of the input signal. In (fig.:8) the pair of peaks is now expanded from 9500Hz to 5500Hz and 13500Hz. The peaks moved thus in each case around 4000Hz, the frequency of the input signal. This connection is up(fig.: 9) to be clearly recognized.
The notch in the spectrum of the noise, with the respective input frequency, results from the notch-filtering.

Test 5
Which influence does the level of the input signal have on the noise?


Test tones:
16000Hz sinetone (level - 3dB) - 44,1kHz/16b/stereo
16000Hz sinetone (level -10dB) - 44,1kHz/16b/stereo
16000Hz sinetone (level -20dB) - 44,1kHz/16b/stereo
16000Hz sinetone (level -30dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)

Level input signal (dB) -3,-10,-20,-30
Level noise (dB) -2,-12,-23,-33

The level of the disturbance decreases approximately equivalent with the level of the input signal. The level of the input signal has apparently no influence on the spectrum of the noise.

Posted Image
fig.: 11 spectrum of the noise for the input signal with a level of
(-3dB-green, -10dB-red, -20dB-blue, -30dB-yellow)

Test 6
Which influence does the betaPlayers Equalizer have on the noise?


Testtone: 16000Hz -Sinuston (level - 3dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)

EQ-settings (betaPlayer):
preamp: neutral (mid)
60Hz: max
170Hz-14kHz: linear decreasing
14kHz: min
16kHz: max

Its obvious, that the EQ has no influence on the spectrum of the noise. The amplification of the 16kHz-band in the EQ results in a light increase of the noiselevel. This test shows, that the noise developes after the audiostream passed the EQ. If the noise would have been developed prior to the EQs influence, their noises spectrum has to be decreasing with increasing frequencies.


Posted Image
fig.: 12 noise spectrum with EQ (green: EQ off)
(blue: EQ on with mentioned settings)


Test 7
Which influence do other smartphone-settings have?


Testtone: 16000Hz -sinetone (level - 3dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)

1. StarUp depressed by renaming \storage\windows\StartUp
no influence

2. Flightmode on/off
no influence

3.Bluetooth on/off
no influence

4. Activesync-connection via USB est./not est.
no influence


Test 8
Which influence has the quantization of the audio-data?


Testtones:
16000Hz -sinetone (level - 3dB) - 44,1kHz/16b/stereo
16000Hz -sinetone (level - 3dB) - 44,1kHz/8b/stereo
16000Hz -sinetone (level- 3dB) - 22,05kHz/8b/stereo
23000Hz -sinetone (level -3dB) - 48,00kHz/16b/stereo (added on 02-15-05)

Testfile format: wav (PCM uncompressed)

There are no differences in results, same length, same intervall and same spectrum. The quantization has no influence.


Test 9
Does the bug also occur in playback via internal speaker?


Testtone: 16000Hz -sinetone (level - 3dB) - 44,1kHz/16b/stereo

Testfile format: wav (PCM uncompressed)

I've started the test with min. volume and increased it evry 20sec.by pressing the side-button. Already at 2/5-volume was the noise audible clearly through the internal speaker.

Conclusion:

Cause of the many articles about this bug in this forum, i would allmost claim, that this bug is detectable at all HTC-Smartphone-modells, where the bug is in at least one case known. I know that some of you would decline this, but i think the facts argue for themselves. People who claim that they don`t have this bug, IMHO simply don't hear it. It took a long time for me to notice the bug too, i've heard many many songs on my mobile without noticing the bug, but then i've heard Genesis "Watcher of the skies" and the bug was more than audible. Since this experience, i started to hear the noise in nearly every song i listened to concentrated. The regularity in the bugs appearance and the many many users who noticed that bug, make it hard to believe that there is any HTC-smartphone without that bug, these can't be isolated cases!
The tests have shown, that the appearance of the bug depends on the frequencies in the played audio-file, thus its clear, that the audibility and the annoy-potential of the bug is strongly related to the kind of prefered music. For instance a piano or string solo with isolated heights should make the bug more audible than a rock-song with fast drums and overdrived e-guitars. Even for "goldears" the bug should be hardly audible in this case. The more complex and broadbandy a song is, the harder the bug should be audible. Furthermore the noise goes under in the music itselves.
© by Bazzo 2005

i hope my translation is not to bad for understanding, sorry for bad english 8-[

best regards Bazzo :lol:

Attached Files


Edited by bazzo, 18 February 2005 - 05:32 PM.

  • 0

#2
pietta

pietta

    Regular

  • Members
  • PipPip
  • 79 posts
HTC technicians need to have a look at this ASAP before orange releases the next upgrade. fingers crossed

Edited by pietta, 14 February 2005 - 10:00 PM.

  • 0

#3
Pondrew

Pondrew

    Hardcore

  • Members
  • PipPipPipPipPipPip
  • 4,404 posts
  • Location:South Croydon
  • Devices:MDA Vario II
You're not really suggesting that my E200 and the original SPV I had before had this bug. Are you?

Afraid your conclusion to me sounds like if you can't hear it then you must be wrong. Strange logic...

  • 0

#4
rct_uk

rct_uk

    Newbie

  • Members
  • Pip
  • 45 posts
  • Location:London, UK
  • Devices:HTC Touch Dual; Orange SF
Thank you for your hard work bazzo! I hope that this will go some way to solving the mystery.

  • 0
"In three words I can sum up everything I've learned about life: It goes on." Robert Frost.

#5
bazzo

bazzo

    Regular

  • Members
  • PipPip
  • 96 posts
  • Devices:SDAmusic

You're not really suggesting that my E200 and the original SPV I had before had this bug. Are you?

<{POST_SNAPBACK}>


ok, my fault i simply messed up the translation, ill cange it instantly. I meant, the bug is on all devices, where at least one single device of the same modell is known to have tha bug. and i know about (SDA, SDAmusic, C500, SP3i) no idea about E200 and SPV I

Afraid your conclusion to me sounds like if you can't hear it then you must be wrong. Strange logic...

<{POST_SNAPBACK}>

yes, its a bit strange... but its my considered opinion that many of the users claiming theire device to be bugless are wrong.

Edited by bazzo, 18 February 2005 - 05:47 PM.

  • 0

#6
spark9990

spark9990

    Regular

  • Members
  • PipPip
  • 52 posts

Afraid your conclusion to me sounds like if you can't hear it then you must be wrong. Strange logic...

<{POST_SNAPBACK}>



i agree with bazzo, if you cant hear it then your wrong (only for the HTC models after e200, basically all the latest HTC models). People have different abilities of hearing, some frequencies might be hearable to them.

  • 0

#7
Pondrew

Pondrew

    Hardcore

  • Members
  • PipPipPipPipPipPip
  • 4,404 posts
  • Location:South Croydon
  • Devices:MDA Vario II

i agree with bazzo,  if you cant hear it then your wrong (only for the HTC models after e200, basically all the latest HTC models).  People have different abilities of hearing, some frequencies might be hearable to them.

<{POST_SNAPBACK}>

And what about the guys (think Disco Stu may have been one of them) whose spvs developed the problem only for it to disappear again. I suppose their ears are playing tricks on them?

Look, I don't doubt there's an issue but I find it difficult to believe it's universal. Are you saying Paul and co are deaf?!

  • 0

#8
Pondrew

Pondrew

    Hardcore

  • Members
  • PipPipPipPipPipPip
  • 4,404 posts
  • Location:South Croydon
  • Devices:MDA Vario II
Anyway, credit where credit due. A good piece of work bazzo. Was this a thesis for your studies?

  • 0

#9
bazzo

bazzo

    Regular

  • Members
  • PipPip
  • 96 posts
  • Devices:SDAmusic

Anyway, credit where credit due. A good piece of work bazzo.

<{POST_SNAPBACK}>

Thanks :(

Was this a thesis for your studies?

<{POST_SNAPBACK}>

no ive made the test cause i love my phone, its the best i ever had, but i want it to play my music in a way i could enjoy, thats all.

And what about the guys (think Disco Stu may have been one of them) whose spvs developed the problem only for it to disappear again. I suppose their ears are playing tricks on them?

Look, I don't doubt there's an issue but I find it difficult to believe it's universal. Are you saying Paul and co are deaf?!

<{POST_SNAPBACK}>


i know a guy in my "home-forum" who had 4 SDAs and 1 SDAmusic, all devices with this bug. I had a SDA with that bug, now i have a SDAmusic with that bug. Belive me, i thougt long about claiming this.

PM Paul an co, and ask them to download the Adobe Auditon Trial or cooledit or anything else and repeat my tests on theire phones, and we'll see.

I'm the first who revises my claims if someone proves that black is white.

i'll go asleep now.
regards bazzo :lol:

  • 0

#10
xenophobe

xenophobe

    Enthusiast

  • Members
  • PipPipPip
  • 264 posts
  • Devices:Audiovox SMT 5600
Wow... excellent work! Do you have a government job? lol (just wondering where you found all the free time.... heh)

Really though.... excellent analysis!

  • 0

#11
niacin65

niacin65

    Regular

  • Members
  • PipPip
  • 142 posts
  • Location:Philippines
  • Devices:HTC Hero, O2 XDA Graphite

i've heard many many songs on my mobile without noticing the bug, but then i've heard Genesis "Watcher of the skies" and the bug was more than audible. best regards Bazzo :lol:

<{POST_SNAPBACK}>



Wow, thanks for the audio test. BTW, I haven't heard that Genesis track in ages. I love it, and saw them perfom it live in Detroit mid 70's.

  • 0
Posted Image

#12
pharma79

pharma79

    Regular

  • Members
  • PipPip
  • 58 posts
Hi BAZZO ,

I have put a link to your work in clubimate`s forum . Hopefully it maybe useful so that developers may take alook at it .

Anyway thanks for hardwork!

  • 0

#13
scott2eyes

scott2eyes

    Diehard

  • Members
  • PipPipPipPip
  • 429 posts
  • Location:London
  • Interests:Star Wars, Stone Roses, Shiny things.
  • Devices:iPhone 3G
Wow... now that's what I call doing your homework!

Good work.

  • 0
My (slightly out of date!) guide to the MPx200 (including customising the homescreen and Media Player):

*LINK*

#14
bazzo

bazzo

    Regular

  • Members
  • PipPip
  • 96 posts
  • Devices:SDAmusic

...and saw them perfom it live in Detroit mid 70's.

<{POST_SNAPBACK}>


i really really envy you for that!!! sry for being OT now 8)
regards Bazzo

  • 0

#15
XanderD

XanderD

    Newbie

  • Members
  • Pip
  • 7 posts
  • Devices:Orange SPV C500
Fantastic!
The last time i went back to the Orange Shop to ask why this *tsss* is there, they basicly told me that i have faulty mp3-files and wma´s and further more that i am insane...

Ill go back this afternoon, and show em!

thnx a lot. i hope they see its real this time, and do something about it.

  • 0

#16
ddieter

ddieter

    Regular

  • Members
  • PipPip
  • 111 posts
  • Location:UK
  • Devices:I-mate JAMin, C500
Excellent work. Now we have proof for all those deaf people who claim no to hear it. Unless of course they are blind as well! :lol:

  • 0

#17
klacik

klacik

    Newbie

  • Members
  • Pip
  • 11 posts
  • Devices:Orange SPV C500

Excellent work.  Now we have proof for all those deaf people who claim no to hear it.  Unless of course they are blind as well!  :(

<{POST_SNAPBACK}>


LOL :lol:

  • 0

#18
Pondrew

Pondrew

    Hardcore

  • Members
  • PipPipPipPipPipPip
  • 4,404 posts
  • Location:South Croydon
  • Devices:MDA Vario II

Excellent work.  Now we have proof for all those deaf people who claim no to hear it.  Unless of course they are blind as well!  :lol:

<{POST_SNAPBACK}>

You have proof of it occuring on SOME devices... you would have to test a batch and then do a test of statistical significance before you could begin to claim to have proof of it happening on MOST/ALL devices.

  • 0

#19
bazzo

bazzo

    Regular

  • Members
  • PipPip
  • 96 posts
  • Devices:SDAmusic

...you would have to test a batch and then do a test of statistical significance before you could begin to claim to have proof of it happening on MOST/ALL devices.

<{POST_SNAPBACK}>


...or we have to wait for HTC to recognise, that they sell devices with an epidemic error...


but u'r right, its only a assumption that its an epidemic error, even if i'm assured

Edited by bazzo, 16 February 2005 - 02:34 PM.

  • 0

#20
a_c

a_c

    Newbie

  • Members
  • Pip
  • 2 posts
  • Devices:SPV C500
Before i do anything stupid with the phone i need some advise! Would it be possible to install the operating system from the original SPV - Smartphone 2002? This way i could see whether the hiss then stops... If so, how would i do this?

It's just a thought...

  • 0




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users