Log in

View Full Version : Menneisyys On A2DP: Widcomm and Microsoft BT Throwdown!


Jon Westfall
11-07-2006, 03:00 PM
<div class='os_post_top_link'><a href='http://www.pocketpcmag.com/blogs/index.php?blog=3&title=the_a2dp_support_aamp_sound_quality_aamp&more=1&c=1&tb=1&pb=1' target='_blank'>http://www.pocketpcmag.com/blogs/in...1&c=1&tb=1&pb=1</a><br /><br /></div><i>"There is not really competition between the A2DP quality of the two Bluetooth stacks. If you happen to have a WM5-upgraded Dell Axim x50(v) or x51(v), don’t hesitate: install the Widcomm BT stack on it if you want the best available sound quality. You’ll be presented definitely better sound quality, on the expense of slightly (but not much!) decreased battery life. And, of course, you get a lot of additional Widcomm goodies like Bluetooth PAN and the like. Similarly, if you need to choose between, say, a Widcomm BT stack-based device and a MS-based one (and there's no known Widcomm hack for the latter), go for the earlier, if, otherwise, the two devices are the same."</i><br /><br />A2DP is a topic near to my heart, as it is something I'm truly happy to have in my devices now (And something I was really annoyed not to have 10 months ago). Strangely, I find that Microsoft's A2DP Profile in AKU 3 devices to be fairly bug free and not as processor intensive as the AKU 2 version that Menneissys has looked at. Any A2DP users out there that want to weigh in - Widcomm or Microsoft BT stack for advanced audio?

pbiba
11-07-2006, 03:35 PM
I've been using my Cingular 3125 with two different models of Logitech headphones (the new FreePulse and the earlier version) and both work perfectly and have excellent sound quality.

Menneisyys
11-07-2006, 03:46 PM
I find that Microsoft's A2DP Profile in AKU 3 devices to be fairly bug free and not as processor intensive as the AKU 2 version that Menneissys has looked at

BTW, what device are you running AKU3 at? I'd really like to know whether the sound quality bug (which can be described as not using a low-pass filter designed to comply with Nyquist–Shannon sampling theorem (http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem) - it's known for anyone that has ever with digitalizing sound) has been fixed in the A2DP support of AKU3 by Microsoft.

Menneisyys
11-07-2006, 03:55 PM
(I've just added a new section on the exact, electric engineer-friendly description of the sound problem of the MS BT stack.)

x51vuser
11-07-2006, 06:36 PM
Do not use Widcom for dell Axim x51v - it is unstable.
A2DP over MS BT works well unless you try to play stream from net over WiFi - that is only bug I found so far and it affects not only Dell's but others as I found reading xda-developers forum

Darius Wey
11-07-2006, 06:46 PM
BTW, what device are you running AKU3 at? I'd really like to know whether the sound quality bug (which can be described as not using a low-pass filter designed to comply with Nyquist–Shannon sampling theorem (http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem) - it's known for anyone that has ever with digitalizing sound) has been fixed in the A2DP support of AKU3 by Microsoft.

I believe Jon's using the T-Mo Dash.

I can also test this out for you when the X500 arrives on my doorstep. That uses AKU3.2.

Don Tolson
11-07-2006, 08:29 PM
I've read through this article and most of the two threads on Aximsite related to this project.

although I found what I believe to be the appopriate download, hat I guess I missed was exactly how to install the Widcomm stack on an Axim X50V WM5.

The discussion is very technical and not for the casual PPC user.

Menneisyys
11-07-2006, 08:39 PM
I can also test this out for you when the X500 arrives on my doorstep. That uses AKU3.2.

That'd be very nice. I'll also try to test the A2DP with a strong low-pass pre-filtering (that is, the A2DP coder will only get a sufficiently low-pass-filtered content and, therefore, there can't be aliasing) so that we can find out whether it's indeed aliasing, or something else.

Menneisyys
11-07-2006, 08:41 PM
although I found what I believe to be the appopriate download, hat I guess I missed was exactly how to install the Widcomm stack on an Axim X50V WM5.

Just execute the CAB file in BroadcomBluetoothStackX51V05.zip - it'll automatically install the Widcomm stack on your device without your need to do anything else. After a reboot, you'll already have Widcomm.

Menneisyys
11-07-2006, 08:45 PM
Do not use Widcom for dell Axim x51v - it is unstable.

It depends on what you're using it for. I've been using it for more than a month, 6-10 hours a day, to feed music into my Pulsar 590A via A2DP - I haven't ever encountered problems.

Of course, it might have problems with, say, some GPS modules, but, for A2DP (and AVRCP), it's excellent.

A2DP over MS BT works well

It does work well - it's just that the sound quality is not so good as with the Widcomm BT stack, under exactly the same circumstances, the same hardware, with all sound sources. This, in addition, isn't just a Dell Axim x51v-specific problem - I've enountered exactly the same behaviour and, comparatively, bad (aliased-like) sound quality on my Wizard. On the latter, of course, I couldn't compare it to the Widcomm BT quality as there aren't known Widcomm BT hacks for the Wizard. (Unfortunately...)

x51vuser
11-07-2006, 09:36 PM
A2DP over MS BT works well

It does work well - it's just that the sound quality is not so good as with the Widcomm BT stack, under exactly the same circumstances, the same hardware, with all sound sources.

I paired Axim with MMV-200 (from SE). Quality excellent for MS BT.
(except when streaming over WiFi but that affects also other PDAs as I said)

Menneisyys
11-07-2006, 10:21 PM
I’ve played a bit with (drastic) low-pass filtering with Pocket Player to find out whether not letting high frequencies (which can cause aliasing) helps the situation.

Nope. It only worked when I set the (drastic) cut frequency to around 800 Hz (that is, I haven’t let anything out of the PDA above 800 MHz). It seems the most affected frequency band is between 800 Hz and 3000 Hz – that is, the frequency band mostly occupied by human sound. This is why I’s mostly with (clean) singing that the very bad distortion can be clearly heard.

In addition to the above-linked / mentioned Värttinä title, I’ve also collected some titles that sound (in cases, much) worse with the MS BT stack than with the Widcomm one (trust me: they DO sound much worse. I’m pretty much into Hi-Fi – I’m not a noob in sound engineering and related areas). Sorry, they’re all Finnish; the case must be the same with English stuff (make sure you listen to some sound with pronounced singing, let it be either female or male). Dunno if they (or demos of them) are available on the Web; if you can’t find them, give a try to the already-linked Värttinä title, it contains enough singing to show you the difference.

Clearly audible difference:
Pihikoppelo – Kultamunat – Kraatari (4th title on the CD), Talon perijä (8th), Siantappojenkka (10th)

Awful with the MS stack:
Pihikoppelo – Kultamunat – Vangin kohtalo (6th) (male singing!)
Jyri Makkonen - Ja pokasaha soi – Palapan Killi (19th)
Maailman Matkaajat - Mielen valtias (7th)

Note that the source format has no effect on the effect – they can even be uncompressed WAV files, the distortion will be heard.

I’d like to point out again that I’m a Hi-Fi / audiophile guy. That is, I notice a lot of things (for example, excess distortion) ordinary people may not notice. My music preferences aren’t in the instrumental-only world either (with them, the distortion will be much less conceivable for the frequency band reasons outlined above. That is, if you only listen to, say, jazz, you may never run into this problem at all). Also, as my BT A2DP headphones (Plantronics Pulsar 590A) are the best (according to most A2DP people), sound quality-wise, with much worse headphones, the quality difference between the two BT stacks may be less pronounced.

Menneisyys
11-07-2006, 10:24 PM
A2DP over MS BT works well

It does work well - it's just that the sound quality is not so good as with the Widcomm BT stack, under exactly the same circumstances, the same hardware, with all sound sources.

I paired Axim with MMV-200 (from SE). Quality excellent for MS BT.
(except when streaming over WiFi but that affects also other PDAs as I said)

If you have some free time, could you compare the sound quality of the two stacks with some songs that have either male or female singers? (I can even send you some of the above-listed test songs in PM.) I'd really like to know how this all sounds on other headphones.

Marcel_Proust
11-08-2006, 01:18 AM
This is a really noobie question...you'll have to excuse me.
But if I"m using Ivan's 3.2 AKU on the HTC Universal, I think it's using the MS stack. And I do find the frequency problem as described.
Is there a way of getting the Widcomm for the Universal?

Ploobers
11-08-2006, 03:24 AM
Is A2DP available for the K-Jam yet?

badbob001
11-08-2006, 05:35 AM
Doesn't replacing the ms bt stack with the widcomm one eat up more of the axim's scarce ram (program memory, not storage memory)?

x51vuser
11-08-2006, 07:32 AM
Sorry Menneisyys but I am not willing to experiment at the moment - maybe at next ROM update I would find some time.
Please PM references or samples of your very interesting collection :lol:
I must also warn I am not that music expert so it could turn I just do not notice differences ....

Menneisyys
11-08-2006, 09:00 AM
This is a really noobie question...you'll have to excuse me.
But if I"m using Ivan's 3.2 AKU on the HTC Universal, I think it's using the MS stack. And I do find the frequency problem as described.
Is there a way of getting the Widcomm for the Universal?

Nope, right now, not (yet?). I'll try to find why the Dell hack doesn't work on the HTC Universal / Wizard. For now, you have to use the MS BT stack.

Menneisyys
11-08-2006, 09:14 AM
Doesn't replacing the ms bt stack with the widcomm one eat up more of the axim's scarce ram (program memory, not storage memory)?

Yes, 4Mbyte when active and 2.5M when inactive, as can be seen in the following chart (original HTML here (http://www.winmobiletech.com/kuvat/x51vmsvsWCBTStackMemUsage.html)) I've made:

http://www.winmobiletech.com/kuvat/x51vmsvsWCBTStackMemUsage.png

In here, I've shown the free dynamic RAM memory on my A12 x51v (no Today plug-ins or other styuff; everything measured after a hard reset) left after, with the Widcomm BT stack, (de)installing and en/disabling the given BT stack.

This, however, isn't that big an issue in most cases - it's very rarely that you'll need the extra 2-4M (for example, when running some memory-intensive games). The advantages of the Widcomm BT stack, at least in the A2DP and the PAN case, are apparent.

Menneisyys
11-08-2006, 09:17 AM
I must also warn I am not that music expert so it could turn I just do not notice differences ....

It heavily depends on the music type. If it' instrumental and the instruments used don't really use the 800 Hz .... 3 kHz frequency band, there won't be really bad distortion.

Of course, the Widcomm BT stack is also considerably better in NOT cutting off the highs.

Menneisyys
11-08-2006, 09:22 AM
I've also given a try to the voice/singing-only Rajaton's songs (http://www.rajaton.org/main.site?set_language=eng). They all suck with the MS BT stack.

XUHAB79
11-08-2006, 09:29 AM
What about X50v users still using WM2003SE? I don't have much time for exploring this kind of thing myself. :?

Menneisyys
11-08-2006, 09:32 AM
What about X50v users still using WM2003SE? I don't have much time for exploring this kind of thing myself. :?

ALL Widcomm versions are OK - I haven't run into any sound quality problems with any of them. (I've tested them all on my several Widcomm-based Pocket PC's - see my previous A2DP tests)

That is, WM2003SE users have just the right BT stack.

Menneisyys
11-08-2006, 09:43 AM
Is A2DP available for the K-Jam yet?

Nope, not officially, only with MS BT stack hacks and custom-cooked ROM versions like that of Molski 2.26.

Menneisyys
11-08-2006, 09:51 AM
BTW, I've thoroughly tested the sound quality of the Palm-based Softick Audio Gateway with the above-mentioned titles. They don't exhibit any kind of distortion either with my T3 + 590A combo. That is, all this is clearly an algorithmic (software) problem in the MS BT stack.

Faenad
11-08-2006, 01:31 PM
Using the HP Stereo Bluetooth Headphones, i noticed a better quality with my Qtek S200 (aka HTC Prophet) (A2DP hack) than with my Hx4700 (Widcomm, with the a2DP Widcomm driver found on aximsite).

Also, after 30-40min of music with the x4700 the music was suddenly becoming scrambled, heavily distorted; I had to power off then power on the headset.
I don't have this issue with the Prophet.

The best quality i have so far was with a 3,5 jack to Bluetooth dongle : Sticked on an excellent iriver MP3 player the quality was top notch.

Menneisyys
11-08-2006, 01:48 PM
Also, after 30-40min of music with the x4700 the music was suddenly becoming scrambled, heavily distorted; I had to power off then power on the headset.
I don't have this issue with the Prophet.

Yup, this (which is because an internal buffer seems to be completely filled in) is a common problem with all the A2DP hacks; the FlowCheck hack can get rid of it. You could also give a try to the official HP 1.6 A2DP Widcomm upgrade on the hx4700. You may have to, however, use the FlowCheck hack I've published - a lot of "hacked" Widcomm devices need it, including the x51v.

Of course, the best solution is the built-in (Widcomm 1.7-based) A2DP support in WM5 version 2.01 on the hx4700 - it needs no hacking.


The best quality i have so far was with a 3,5 jack to Bluetooth dongle : Sticked on an excellent iriver MP3 player the quality was top notch.

Thanks for the report! With the A2DP dongle that comes with the Plantronics 590A, I don't find any quality difference between the (Widcomm 1.7-based) A2DP support in WM5 version 2.01 on the hx4700 (or the x51v) / the Palm-based Softick Audio Gateway and the A2DP dongle.

Good to know the situation is different with some other headphones. Probabloy because they, internally, don't use A2DP (but pfor example physically an entirely different, for example even radio-based protocol using much lower frequencies etc)

Faenad
11-08-2006, 05:43 PM
You may have to, however, use the FlowCheck hack I've published

Thanks for the tip (and for publishing it) !

Thanks for the report! With the A2DP dongle that comes with the Plantronics 590A, I don't find any quality difference between the (Widcomm 1.7-based) A2DP support in WM5 version 2.01 on the hx4700 (or the x51v) / the Palm-based Softick Audio Gateway and the A2DP dongle.


In fact the difference in quality I noticed come from the iriver player.
It have a ''3D sound rendering" effect which works great. The good news is that this effect is conserved when using the 3.5 jack to bluetooth dongle on it : Bluetooth conversion don't alter sound quality, and the HP headphones are up to the task as well. You just lose controls via AVCRP.

To be consistant I should have to test the Bluetooth Dongle on an A2DP device, but I cannot for the moment as my S200 only has a 2.5 jack.

If I have time I will test the dongle on my 2003SE Hx4700 and compare it with A2DP with official HP 1.6 drivers + your hack.

Don Tolson
11-09-2006, 12:23 AM
Just execute the CAB file in BroadcomBluetoothStackX51V05.zip - it'll automatically install the Widcomm stack on your device without your need to do anything else. After a reboot, you'll already have Widcomm.[/quote]

OK, I got the V03 version. I'll try it with the V05. thanks.

Menneisyys
11-09-2006, 08:20 AM
BTW, I've tested the iPAQ hw6915 (Widcomm 1.7.1). Works great (equally good as the x51v with the Widcomm stack + FlowControl hack or the WM5-upgraded hx4700), no problems at all.

Videonut
11-18-2006, 06:22 PM
That is, all this is clearly an algorithmic (software) problem in the MS BT stack.

I too have been experiencing this distortion problem on my UTStarcom(Apache) 6700 w/MS BTStack. I had the Plantronics 590a which has amazing sound quality but distortion. I now have the Lubix Nc1 which also has great sound quality and distortion. However I had tried the Moto HP820s and they didn't seem to have the distortion problem but the sound quality wasn't great and clipping would occur at times on highs and bass, leading me to believe that had the Motos had better speakers the distortion would have occured. Now why does no one seem to have distortion with the Jabra 620s? Is the sound quality lower than that of the Plantronics or Lubix or could it be that their algorithm for turning the samples to music is better? I would love to hear from someone who has compared headsets side by side.

After reading the Nyquist Sampling article I was wondering if you(Menneisyys) had also tried running your samples through a highpass filter and then low pass filter? All tests I have done were to increase sampling rates and trying different encoders like Lame for MP3s and Ogg Vorbis at very high bit rates. Maybe I'll give a try at lower bit rates, but I'm starting to feel that no matter what rate the aliasing will occur because the MS algorithm will screw up whatever we throw at it. Therefore the only solution would be an advanced coderer hacking the MS Stack or a port of the Widcomm Stack.

Menneisyys
11-18-2006, 06:50 PM
That is, all this is clearly an algorithmic (software) problem in the MS BT stack.

I too have been experiencing this distortion problem on my UTStarcom(Apache) 6700 w/MS BTStack. I had the Plantronics 590a which has amazing sound quality but distortion. I now have the Lubix Nc1 which also has great sound quality and distortion.

Thanks for the report - that is, this problem is a generic one and not just related to the 590A.

Now why does no one seem to have distortion with the Jabra 620s? Is the sound quality lower than that of the Plantronics or Lubix or could it be that their algorithm for turning the samples to music is better?

The same we've disucssed over HowardForums. So far, noone knows why the Jabra delivers better sound quality (or, to put in another way, less? no? distortion) with the MS BT stack.


After reading the Nyquist Sampling article I was wondering if you(Menneisyys) had also tried running your samples through a highpass filter and then low pass filter? All tests I have done were to increase sampling rates and trying different encoders like Lame for MP3s and Ogg Vorbis at very high bit rates. Maybe I'll give a try at lower bit rates, but I'm starting to feel that no matter what rate the aliasing will occur because the MS algorithm will screw up whatever we throw at it. Therefore the only solution would be an advanced coderer hacking the MS Stack or a port of the Widcomm Stack.

Yup, I've tried very hard to find out whether it's a low-pass filter-related problem. Nope - the MS BT stack distortion is clearly perceivable with anything sung, even if you use a, say, 2 kHz filtering frequency (effectively cutting everything above 2 kHz).

joe_dude
11-22-2006, 05:41 AM
Hi there,

Just curious, has anyone compared the AKU3 vs. AKU2 MS BT stacks? Especially for sound quality?

I'm running the A2DP patch on the Prophet, so no need to mention its poor BT sound quality. :wink: