Better OSM feature recording - a thought

I capture my highway driving to a BlackView, then locally process/geoencode/choose images for entering features into OSM on the same day. I upload the vids to Mapillary perhaps a month later.

I have been frustrated of late by the 4K wide angle imagery not being good enough to read some road signs, rural mailbox names etc leaving many as fix-me’s on OSM. I just started looking at the audio recording capabilities of the camera and now process these to wavs that I can play in sync with the images as needed.

The mp4’s are all date/time stamped so I run a script to extract the wav, then use sox to filter out the road noise.

mkdir wav
for F2C in ls -1 *mp4|cut -c 1-15
ffmpeg -i $F2C_NF.mp4 wav/$F2C.wav

cd wav
mkdir …/hpf
for F2C in ls -1 202*.wav ; do
sox $F2C …/hpf/$F2C sinc -1 300 gain -n

The rate of road signs is relatively low for me, perhaps only 1-2 per minute maximum, so this won’t suit high density driving. My only additional action is wave my finger near the camera left side (starts audio recording) then voice the sign details in the direction of the camera! It seems the road noise is sufficient to keep the recording active. (Permanently on does seem configurable though)

(Note the single command quotes are interpreted as formatting by this forum)