31st January 2013
|
#1 | | Gear maniac
Joined: Apr 2011 Location: Minneapolis, MN
Posts: 175
Thread Starter | GoPro pseudo-analysis
I spend a lot of my time referencing things that were shot via gopros, which is unfortunate since the sound quality isn't stellar. Gopros have a very distinct sonic quality, but after many hours of searching around I wasn't ever able to find an analytical look at why they sound the way they do, or more importantly if it could be corrected.
So I took my HD1 and my Fuzzmeasure RTA rig and did some quick, informal tests. Here's the results:   
interesting notes:
1. Huge cutoff at 5.5K (probably not repairable via eq)
2. Resonant series at 1.5, 2.4, 4,0, and 5.1 KHz, most likely related to the case/mic.
I'd really like to revisit this with the newer HD2 and HD3 models, as well as test the variety of open-back cases to see how things change. I also would really like to get my hands on an HD2 or HD3 with the "mic in" ports to do an isolated electronics test to see what limitations exist in the acoustical hardware (mic and case) versus the processing and encoding.
| 
|  | |
31st January 2013
|
#2 | | Gear maniac
Joined: Jan 2011
Posts: 272
|
Looks like th gopro mic has a low pass set around 7k. Wow that sucks.
| |
| | |
31st January 2013
|
#3 | | Gear addict
Joined: Aug 2007  |
I've always known the Go Pro to have horrible aliasing issues, any analysis on that?
| |
| | |
31st January 2013
|
#4 | | Gear maniac
Joined: Apr 2011 Location: Minneapolis, MN
Posts: 175
Thread Starter |
^ I'm not familiar with a test procedure for that, any suggestions?
Keep in mind too that this is the HD1, which is now 2 model generations old. I don't believe the HD2 had any significant upgrades, but the HD3 touts significant improvement. I'm trying to see if there's one around my workplace to borrow.
| |
| | |
31st January 2013
|
#5 | | Lives for gear
Joined: Jan 2007 Location: Barcelona
Posts: 713
| Quote:
Originally Posted by Chippy569  ^ I'm not familiar with a test procedure for that, any suggestions? | Record a pure tone at healthy level at high-ish frequency. If you look at the recorded signal you should see that same pure tone peak obviously, but aliasing artifacts also introduce extra peaks at certain alias frequencies.
1/2 sampling frequency acts like a mirror where frequencies get reflected.
| 
| | |
31st January 2013
|
#6 | | Gear maniac
Joined: Apr 2011 Location: Minneapolis, MN
Posts: 175
Thread Starter |
Awesome, I can definitely run that through. Would a screenshot of RX (or similar) be sufficient evidence?
Also, I've landed access to a Hero3 for tomorrow (hopefully) so you can look forward to more thorough analysis around then.
| |
| | |
31st January 2013
|
#7 | | Lives for gear
Joined: Apr 2009
Posts: 1,269
|
Out of curiosity:
What's the point of doing all this? Isn't it like finding scientific proof for that music sounds crap when listened over a mobile phone speaker?
Proving what's obvious: The mic of a little plastic camera sounds crap?  | |
| | |
31st January 2013
|
#8 | | Lives for gear
Joined: Jan 2007 Location: Barcelona
Posts: 713
| Quote:
Originally Posted by apple-q Out of curiosity:
What's the point of doing all this? Isn't it like finding scientific proof for that music sounds crap when listened over a mobile phone speaker?
Proving what's obvious: The mic of a little plastic camera sounds crap? | Yes, but the GoPro are known to have a certain kind of crappy sound. It's like a design flaw in the ADC circuits that could have been avoided. I guess it's also out of curiosity to have a graphic evidence of what's happening exactly.
| |
| | |
31st January 2013
|
#9 | | Lives for gear
Joined: May 2009 Location: atlanta
Posts: 1,777
|
it is going to depend on the case its in.. there is a skeleton case and one that is open on the back ony and a fully water proof one.. I have use the audio at times with skeleton case.. takes a little EQ to get it sounding acceptable. ,,not great. but usable .
| |
| | |
31st January 2013
|
#10 | | Lives for gear
Joined: Feb 2009 Location: Brussels
Posts: 505
|
I would avoid using GoPro sound at all times.
It just sounds horrible.
The problem with their sound is not its frequency response, which you could treat thanks to your findings. But it is, as someone already said, its awful aliasing sound, which I haven't found a cure for as of yet.
| |
| | |
1st February 2013
|
#11 | | Gear maniac
Joined: Apr 2011 Location: Minneapolis, MN
Posts: 175
Thread Starter | Quote:
Originally Posted by apple-q Out of curiosity:
What's the point of doing all this? Isn't it like finding scientific proof for that music sounds crap when listened over a mobile phone speaker?
Proving what's obvious: The mic of a little plastic camera sounds crap? | The point isn't to say "wow, that's terrible!" -- it's to determine why, or in what way, it is terrible in order to determine if it is possible to reduce the terrible-ness in a simple way.
To that end, I have a new graph to show, this is a pair of "correlations" - think of it as reference mic minus gopro results. I also ran a frequency analysis on my HD1 uncased, which is in olive drab: 
As you can see, there is a 3dB drop right at 1k with the case on, but overall the results are surprisingly similar.
| |
| | |
1st February 2013
|
#12 | | Lives for gear
Joined: Feb 2009 Location: Brussels
Posts: 505
|
I realize and admire you're doing this for the better cause, but analysing the frequency response won't get you far.
The problem with gopro's is far worse. It's the artifacts of the poor conversion that are most audible.
The fact that the mic is rolled off at 6k isn't so bad, or if it has peaks somewhere.
The answer to 'Why does it sound' bad is that it has very, very crappy electronics in there for the sound part of it, to make as much room as possible to get more video electronics in that small box.
If you go out with a GoPro strapped to your helmet, just strap a handheld recorder (even the cheapest) to the side of it and you'll ssee a major increase in quality.
| |
| | |
2nd February 2013
|
#13 | | Gear maniac
Joined: Apr 2011 Location: Minneapolis, MN
Posts: 175
Thread Starter |
I have some aliasing results for you. I first ran the test at 10 KHz, but heard, well, nothing because of the low pass at 7-ish K. So I re-ran at 6 KHz. 
Low frequency stuff is likely room noise and etc, but I'm not seeing any clear or distinct aliasing artifacts in here? This test would probably be more definitive with an HD2 or HD3 where the "mic in" port could be utilized to override the acoustical shortcomings.
| |
| | |
6th February 2013
|
#14 | | Gear addict
Joined: Oct 2007 Location: Charlotte, NC
Posts: 443
| Quote:
Originally Posted by Jamie Mac The answer to 'Why does it sound' bad is that it has very, very crappy electronics in there for the sound part of it, to make as much room as possible to get more video electronics in that small box.
If you go out with a GoPro strapped to your helmet, just strap a handheld recorder (even the cheapest) to the side of it and you'll ssee a major increase in quality. | It sounds more like a design flaw than just crappy components. I find it hard to believe there are A/D converters that bad.
__________________
~Will
| |
| | | |
New Reply 
Facebook
Twitter
LinkedIn
