Drone.vc – Request for Companies #1: Better Audio | Hacker News Reader

Drone.vc – Request for Companies #1: Better Audio

dronevc.tumblr.com

40 pointsdweekly27 comments

This isn't really a problem particular to drones. Most audio applications have noise issues - room reverb, wind noise, background chatter, preamp noise etc.

A couple of options spring to mind:

1) Get a noise fingerprint from the rotors and clean the audio with DSP. Hard because the noise of the rotors is constantly changing, but you could potentially place several contact mics next to the rotors and then subtract that from the main mic to get a cleaner signal

2) super directional audio with parabolic reflectors which reject audio outside the narrow beam

3) Don't record audio from the drone! Use wireless mics on the ground instead

Disclaimer: Used to be an audio engineer in past life and have been running a drone company for 3 years. We specialise in high end video (RED Epic), but are also working on some non-cinematography applications at the moment. We have both hardware and software teams. Any investors interested in hearing more, you can get in touch at info@londonhelicam.co.uk

It sounds like you're a perfect fit to solve this problem, if you were to choose to do so :-)

Dharma,

Great points!

#1 - my thought was someone very clever could use the ESC output to the motors to estimate likely noise output from that rotor as well as measuring the actual noise output.

#2 I'm genuinely curious about the pros and cons of flying with a parabolic vs an array mic approach. It sounds like you could offer some insights here on a tradeoff.

#3 Well, yes. [grin] But that doesn't exactly solve the problem asked for.

I'll drop you an email now - thanks for responding! ^_^

Interesting David, myself and some of the guys at the office have been experimenting with quads lately and discussing possible military/police or media uses, after reading your initial proposal #1 struck me as a possible solution. However I've since swapped to another much more "back to basic principles" idea that while more difficult, would have a much wider impact.

Just to make sure I understand the problem ... wouldn't the simplest solution be a 5m audio cable hanging from the drone with a wind-shielded microphone at the bottom? I realize there are all kinds of dynamic problems with a setup like this, but 10dB of noise supression could be achieved easily like that.

What about a set of software tuneable band reject filters instead of active noise reduction?

rotor noise tends to be really wide band, simple filtering won't be able to remove it

Genuinely curious, let's assume this technology is created. What would audio signals would you even record?

Most drones are far from any subject (sound-producing or not). What are use cases where the drone is close to a sound source?

This got me thinking of a few examples:

  - drones could get closer to waterfalls or volcanoes than any sound engineer with a boom could
  - a drone could be used for video interviews, automatically panning between the interviewer & the interviewee, recording both video and pristine audio, streaming via a cellular data link back to the studio

I think just subtracting out the rotors would be nice...

Right, but my point is that if a drone is flying over an area like the Golden Gate Bridge, you're really not going to pick up any audio aside from wind noise. You'd have to be in close proximity to some sound producing subject.

Search and rescue missions? People could tell the drone operator how to help.

Not to mention tell the operator if they are alive at all. It may not be obvious if you see from a few meters away someone who is mostly immobilized by a fallen whatever. For this sort of application, a microphone on a winched cable might be useful (n.b.: make a feature to jettison the cable if it gets stuck!).

Yes, strap a megaphone on here and now you've got a flying telephone for search & rescue as well as a chance for an aerial director to talk with folks in the ground without the need to set up a second comms channel (e.g. RF radios).

Action sports are probably the big one. Filming things like MBX or snow sports can be greatly improved by including natural ambient sounds. Tricky to do with a quadcopter.

Drone.vc, request for companies, and the first problem posted is "better audio". Seriously? No one is working on it because there are approximately a million and one drone related problems that are actually worth solving!

Solve the perching problem, or the loss of control channel problem, or better image tracking or make them quieter or smaller or bigger or faster or can fly higher or can carry heavier payload or...

But what do I know? Clearly it is the sound that is holding back drone companies from delivering VC level returns.

word!

There are a lot of interesting problems worth solving in the industry. That's what makes it so exciting! What I'm going to do with this sequence is try and point out a set of problems that are interesting that aren't being solved. It could be that nobody is working on them because all of my ideas are all really awful, but the hope is that some people may get inspired to tackle some new problems in the space. I'm not going to claim that audio recording is the singularly most important problem to solve (you seem to be implying that's what I was getting at) but it is an interesting/common one and I hadn't seen anyone working on it. There are another dozen of these tee'd up, some of which address the other problems you've listed. It sounds like you have some good opinions about what problems need to be solved! I'd love to see a response with a little bit more meat than an ellipsis.

One of the _best_ responses to these will be to find companies already working on these problems, just in secret. For instance, I had someone on Twitter let me know that a company in Hilo, Hawai'i called DroneFlow is actually working on some of these audio issues. I'll be reaching out. :)

BTW, researching perching led me to this beautiful thesis - http://groups.csail.mit.edu/locomotion/perching_media/CoryTh... - what's your take on this contribution to the space?

Forgive me if I'm missing something, it just looks like a single page Tumblr account with a quick blurb about being a "Drone VC", with no details about the people or firm behind it. Does anyone know who's behind "Drone VC" or have any more information on it?

Links to David E. Weekly's syndicate page on Angel List (https://angel.co/dweekly/syndicate/) who appears to be @dweekyl (https://twitter.com/dweekly).

Yessir!

(raises hand) Hi! Correct that it's an AngelList syndicate that I started and as seems to be the world's first drone fund. Coverage @ http://techcrunch.com/2014/06/11/drone-vc-is-a-vc-fund-for-f...

Also: https://twitter.com/dronevc https://www.facebook.com/DroneVC etc ^_^

In other news, I clearly need a better homepage than an AngelList redirect!

Although this approach is technically awesome, can't someone just modify existing noise cancellation technology with mics closer to the motors / props? It's a horrible hack, and nowhere near as elegant, but it may work quite well considering that no matter what signal is used for cancellation, it will be difficult to discriminate wanted / unwanted sounds in the cancelled frequencies.

It's quite possible that this "horrible hack" might be good enough! I'd love to see people experimenting and coming up with answers about what works and doesn't. :)

http://www.eeweb.com/blog/extreme_circuits/voice-bandwidth-f...

It's surprising that intelligence agencies don't have a proper solution already.

Wouldn't the target notice if you flew a drone close enough to them to hear them?

Would they tell anybody if they did?

Reply on news.ycombinator.com