Analysis of the HTC-Soundbug (tsss-noise)

[Guest bazzo]

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)

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.

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

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)

Fig.: 4 spectrum of the noise at 250Hz

Fig.: 5 spectrum of the noise at 500Hz

Fig.: 6 spectrum of the noise at 1000Hz

Fig.: 7 spectrum of the noise at 2000Hz

Fig.: 8 spectrum of the noise at 4000Hz

Fig.: 9 spectrum of the noise at 8000Hz

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.

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.

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:

HTC_Soundbug_isolated_.rar

Edited February 18, 2005 by bazzo

[Guest pietta]

HTC technicians need to have a look at this ASAP before orange releases the next upgrade. fingers crossed

Edited February 14, 2005 by pietta

[Guest Pondrew]

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...

[Guest rct_uk]

Thank you for your hard work bazzo! I hope that this will go some way to solving the mystery.

[Guest bazzo]

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 February 18, 2005 by bazzo

[Guest spark9990]

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.

[Guest Pondrew]

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?!

[Guest Pondrew]

Anyway, credit where credit due. A good piece of work bazzo. Was this a thesis for your studies?

[Guest bazzo]

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:

[Guest xenophobe]

Wow... excellent work! Do you have a government job? lol (just wondering where you found all the free time.... heh)

Really though.... excellent analysis!

[Guest niacin65]

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.

[Guest pharma79]

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!

[Guest scott2eyes]

Wow... now that's what I call doing your homework!

Good work.

[Guest bazzo]

...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

[Guest XanderD]

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.

[Guest ddieter]

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:

[Guest klacik]

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:

[Guest Pondrew]

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.

[Guest bazzo]

...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 February 16, 2005 by bazzo

[Guest a_c]

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...

[Guest Griffen]

Y'know what someone should really do is find a track where the noise is bad and capture it on a PC sound card line so anyone who hasn't heard it can hear it, if you know what I mean. :lol:

I do occasionally notice the noise but its only on a few specific tracks and on a lot of my videos.

[Guest bazzo]

Y'know what someone should really do is find a track where the noise is bad and capture it on a PC sound card line so anyone who hasn't heard it can hear it, if you know what I mean. :lol:

I do occasionally notice the noise but its only on a few specific tracks and on a lot of my videos.

<{POST_SNAPBACK}>

i attached one of my testfiles after notchfiltering to the 1st posting. You will hear the noise and nothing else. This is not a testfile to test your phone, the noise is already in the recording.

The ultimate song to hear the noise with real music is Genesis "Watcher of the skies" so take a look at mummys cd-collection and rip this song. I've found no other song, were the noise is that audible.

regards bazzo

[Guest Griffen]

i attached one of my testfiles after notchfiltering to the 1st posting. You will hear the noise and nothing else. This is not a testfile to test your phone, the noise is already in the recording.

The ultimate song to hear the noise with real music is Genesis "Watcher of the skies" so take a look at mummys cd-collection and rip this song. I've found no other song, were the noise is that audible.

regards bazzo

<{POST_SNAPBACK}>

Thanks, but the file dosen't seem to work, I don't know if its just me but It dosent download!

[Guest bazzo]

hmm, i have no problems downloading, and playing it

regards bazzo

[Guest mcMike]

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}>

Nice work dude.

I just yesterday purchased Qtek 8010 and it has this exact sound issue. I checked as much that the frequency of the occurence is same.

Just when I was wondering is it my sample that is faulty...

Thanks. This is valuable information. Not that I can do anything with it but still.