First round of audio tests done


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,986
Age
46
Location
Ingolstadt
I just finished the first audio tests today, comparing the audio output of a semi-professional USB Interface (FocusRite Scarlett), a Pandora and the TWL6040 on the PandaBoard (which is the same we plan to use for the Pyra).

The test I wanted to do today was testing the audio output of the DAC, not the quality of the amplifier (which will be tested with a different test).

I'vee created an Audio file (48kHz, Uncompressed, 32bit, Mono) has been played back with a PC (FocusRite), a Pandora and a PandaBoard (also uses the TWL6040).

Each device was then being recorded (with another FocusRite Scarlett) and analyzed with Adobe Soundbooth (just a quick analysis).

The result was quite surprising:
I expected the TWL6040 to be the worst one, but actually, the TWL6040 DAC seems to be a bit better than Pandoras one - except when it comes to noise: Then the Pandora wins of all devices.

I guess the bad audio quality with these days devices is the amplifier, not the DAC. But that will be tested as well.

Well, and it seems DACs have become better within the last few years.

You can download the full report with graphics from here: http://evildragon.openpandora.org/AudioTest.pdf

And here (http://evildragon.openpandora.org/AudioTest.zip) is a set including the following:

1. Original WAV-File (that was played back)

2. Each recording from every device (compare the playback quality yourself)

3. Three small WAVs where the audio device is switched while audio is playing (see AudioTest-Report for more information).

Questions? Requests?

Let me know here :)
 
The Pandora does have some pretty nice sound. It doesn't ever crackle or pop or make noises that don't sound right. Even with headphones it sounds pretty nice.
 
Yes, judging by the reduced noise coming from the TV-cable line out of my pandora via my amp and a stupidly long headphone cable compared to plugging my headphones directly into the pandora, at least half the noise of the Pandora's headphone output comes from the headphone amp, not the DAC.
 
The result was quite surprising:

I expected the TWL6040 to be the worst one, but actually, the TWL6040 DAC seems to be a bit better than Pandoras one - except when it comes to noise: Then the Pandora wins of all devices.

I guess the bad audio quality with these days devices is the amplifier, not the DAC. But that will be tested as well.

Well, and it seems DACs have become better within the last few years.
For me it is not such a big surprise since there was Nokia (sigh) behind the development of these chips (but they did not use them). So I would never have expected crap.Regarding noise: I am not sure if the PandaBoard has followed all design rules for lowest noise power supply since it is an Eval board and not necessarily the best possible design. There might also be noise on the clock lines. It is difficult to pinpoint the real reason.
 
Last edited by a moderator:
Pick the best audio hardware for the job remember if it isn't loud enough it can always be amplified 3rd party clean signals are more important that anything else. I remember someone posted here about using xyz name brand headphones and said the quality of the audio on the pandora was awesome. I don't want to be that guy who does not want to do the research and pick whats presented to me. On a selfish level with respect to the task at hand given ED's experience in this field just let him do his thing.
 
Thanks for the report. I'm wondering how much of the subjective quality of the Pandora output is actually due to it having measurably lower fidelity than the others but with a pretty harmonic spectrum.


Do you know the characteristics of the low pass filters used in these systems? How much of the differences in volume at high frequencies is due to the cut off frequency/quality of these filters?


edit: looking at the results again, it's probably all due to the amplifier, isn't it? Looking forward to the next test round.
 
Last edited by a moderator:
One thing about audio playback on pandora. When i use mplayer to playback files, i get a "pop" at the beginning of every track.

Otherwise, aside from a slight hum from the power supply, the sound is clean.
 
I'vee created an Audio file (48kHz, Uncompressed, 32bit, Mono) has been played back with a PC (FocusRite), a Pandora and a PandaBoard (also uses the TWL6040).
Why 32 bits ? I thought everyone (except the hipsters on crack) was pretty much aligned that 16 bits was far enough for any human ear, 24 bits nonsensical except for production recordings ? 
 
Pandora can't do 32bit at all, you were testing ALSA's dmix sample conversion that by default changes to 16bit before sending to hardware..

I'm not sure TWL can do it either, if I understand the spec right it's 16bit only (pandora's DAC at least does 24bit if you bypass dmix).
 
Last edited by a moderator:
For me it is not such a big surprise since there was Nokia (sigh) behind the development of these chips (but they did not use them).
Source? From what I know Nokia were involved a lot with driver development for TI stuff, but I see no indication that they were designing the chips themselves.
 
Last edited by a moderator:
Nice report, I expect much from the audio quality of the Pyra, I'm glad that you are already testing that.The audioquality will be a big part in the Pyra awesomeness. I hope the sound will not be too analytic but rather natural and warm.

What is fixed in the audio part? I guess it's the circuit on the main board, how much this fixed part influence the final audio quality?
 
For me it is not such a big surprise since there was Nokia (sigh) behind the development of these chips (but they did not use them).
Source? From what I know Nokia were involved a lot with driver development for TI stuff, but I see no indication that they were designing the chips themselves.
This statement is based on "historic market insider" knowledge. Nokia and TI worked in a close relationship (strategic alliance) since the 1990ies. The role of Nokia was more to specify the chips (e.g. features, quality) and TI to really develop and produce the silicon by licensing an ARM core and combining with TMS320 DSP. You can see that for example the OMAP3430 was used in Nokia devices before TI brought a "Public" OMAP3530 (which is e.g. used in the OpenPandora). And at that time Nokia was the biggest mobile phone manufacturer - and therefore obviously the lead customer of TI.If you need an external reference about that alliance, e.g. here: http://www.rethink-wireless.com/2011/2/9/OMAP-5-gives-TI-strength-for-life-after-Nokia.htm

A side result is of course that with the introduction of the OMAP devices N770 they became visible in kernel/driver development. But Nokia has a long history in using TI chips without Linux. BTW, the OMAP1 already did have an audio companion chip: TSC2301 (introduced 2002) so that I also can assume several years of growing experience and improvement in the OMAP audio systems.
 
Last edited by a moderator:
The Pandora does have some pretty nice sound. It doesn't ever crackle or pop or make noises that don't sound right. Even with headphones it sounds pretty nice.
In my case I always had some trouble with the Pandora Audio Quality. Most music cause some ugly "hiss" like exposed "ZZZZZ" or "SSSSS", i have to compensate with Equalizer adjustments if the audio player offers one. Seems the Pandora produces more high sounds than lower bass sounds.  I know this also could be because of the Sound file or the earphones but I tested the music files onto my PC (all sounds fine and I only have onboard sound) also tested various Headphones, best sound I have with an huge Sennheiser (of course) compared to my small phillips ear phones.

My ears are pretty sensitive to music, I can hear if a song sounds "wrong".  Overall, the Pandora has good audio, very noticable in games. But I hope there will come a better quality for Pyra AND an standard Equalizer to adjust the high and low audio sounds if needed.
 
Last edited by a moderator:
Back
Top