[OSM-dev] Fwd: Mobile Phone Audio Mapping Tool?

[Graham Jones (Physics)]

David,
Thanks for the comprehensive reply!

My idea is similar to your option (b).  What I envisage doing is pressing a
button on the phone which then records an audio clip, and associates the GPS
location with that clip.  These will be saved on the phone's memory.
It will then be necessary to export the clips and locations to JOSM - the
simplest answer sounds like the use of a GPX file with references to the
audio clips - it will effectively be like setting a waypoint on a Garmin GPS
device, but without the trouble of trying to type in the name with the silly
little joystick while avoiding traffic....  Therefore if you could point me
to what the <link> specification needs to be for JOSM, that would be very
useful please.

The phone creates the clips in 'amr' format, but it shouldn't be difficult
to post-process these into WAV, either as a separate application or as part
of a JOSM plugin.  I'll not try to do that on the phone just yet.

I'll try to get the storage bit working, then should be able to publish
something for people to try....


Graham.

2009/3/27 David Earl <david at frankieandshadow.com>

> On 27/03/2009 21:21, Graham Jones (Physics) wrote:
>
>> I am after a simple way of recording street names when cycling rather than
>> having to write them down.  There is a section on the Wiki on "Audio
>> Mapping", but there is no mention of using a normal mobile phone and
>> bluetooth GPS receiver - the idea is to record audio clips and associated
>> location to import into JOSM later.  The advantage of a phone over the other
>> methods is that it should be able to record both audio, and GPS location at
>> the same time, to avoid the synchronisation problems.  It is also nice and
>> small, and you don't look odd talking into one.   It might even be possible
>> to use a bluetooth headset which would be even easier for cycling...
>>
>> I have had a bit of a play and I think I can make it work (the GPS
>> receiver can talk to the phone, and the phone can record sounds), but I
>> haven't stitched it together in to a single application yet. I can't help
>> but think that someone will have tried this before (and presumably failed
>> because I can't find it anywhere?) - does anyone know if I am onto a loser
>> before I go too far down the road of coding it?  There is a bit more
>> information at
>> http://wiki.openstreetmap.org/wiki/Talk:Audio_mapping#Mobile_Phone_Version.3F
>>
>> Any pointers to where this is going to go wrong would be appreciated!
>>
>>
> Hi Graham,
>
> I wrote the audio bits in JOSM, so I can tell you what goes on at that end,
> and I also added a new bit recently to support something similar to this (so
> make sure you have a JOSM from the last couple of weeks - I'd have to check
> for the exact build number). Incidentally, the help in JOSM is more
> comprehensive than the help OSM wiki page.
>
> However, a word of caution: the fact they are on the same device doesn't
> necessarily mean you can dispense with sync or calibration. It is quite
> likely the GPS track will generate its time stamps from the GPS satellite
> time and the audio manage its sampling and time stamps from the clock on the
> phone. They may not correspond exactly (indeed you probably have to set the
> one on the phone explicitly) and may drift a small amount. I think you
> should check this first. You may be lucky, but don't take it for granted. An
> error of only second or two can make quite a big difference and even if you
> can set the phone clock to the same as the GPS a drift of only 0.1% can make
> 100m difference on a long continuous recording.
>
> How to proceed depends on whether you're recording a continuous sound track
> or a set of snippets.
>
> In all cases you need WAV files. If your phone records another format
> you'll need to convert them to WAV (e.g. using Audacity).
>
> If many, short files then there are two approaches
>
> (a) have a set of WAV files whose (modified) time stamps are those of the
> original recordings (i.e. if you converted them, you'll need to use touch to
> bring the timestamps into sync with the original files).
>
> Tick the appropriate box in the audio preferences to say this is how you
> want to work.
>
> Then when you import audio for a track choose _all_ the files (using shift
> and ctrl click as appropriate in the file chooser dialog). That will
> associate the audio clips at those time points along the track. NOTE: the
> time stamps are those at the *end* of each recording (for obvious reasons, I
> hope), but JOSM will take this into account, so the point identified on the
> track will be at the *beginning* of the recording.
>
> (b) postprocess the GPX file to add <link> elements to the relevant
> trackpoints to refer to audio files. This is a manual equivalent of the
> above. If you want to try this, let me know and I'll look up the exact spec
> of the <link> tag for you.
>
> For a continuous recording, you would still need to sync, because it is
> unlikely your track and recording will start at the same moment. However, if
> you start the recording at the same moment you make a waypoint, you won't
> need to record anything to mark the sync, it will just be precisely the
> beginning of the recording, so you can just briefly start playing and then
> immediately stop the audio from the first waymark and sync. (Actually that's
> true anyway, you could do this on any recorder).
>
> HTH
>
> David
>



-- 
Dr. Graham Jones
Hartlepool, UK
email: grahamjones139 at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20090328/59a666d8/attachment.html>