Audio capture/analysis questions for a small, ~12 hr study

[David Dearing]

Hi Oren and everyone,

I am considering using OpenBadges for a research study focusing on small team interactions.  Our research team has used sociometric badges in the past, but those badges are old and failing so I am currently looking into producing roughly 10-15 badges based on the information in this thread: https://groups.google.com/g/rhythm-badges/c/ISrm1hX-cLE 

I have a few questions I was hoping to get a quick response to before pulling the trigger on production:

• How long will these badges last on a single CR2032 battery?  We are looking to capture data for 12 hour "shifts."  Will these be able to handle that?
  

• Each "shift" of our study will only have 4-5 badges actively collecting data, but those badges will be distributed over a large space with many separate rooms.  How much data can they store before needing to connect to a hub?  Will a single hub (no server) be sufficient if the badges come in contact throughout the day?

• Our interest in these badges is primarily for capturing audio features (we have other means of tracking location data).  There will be more people talking and interacting during the study than just the 4-5 individuals that are wearing a badge.  Is it reasonable to expect the audio data to be useful?  For example, we would like to see:
  • Individual speech features for each individual (speaking, pitch, volume)
    
  • Identify conversations between those wearing badges
    
  • Conversational measures (interruptions, turn taking)  I did see this thread about turn taking, is that still current? https://groups.google.com/g/rhythm-badges/c/SPqijnJp-Ok/m/r9qIDRGICAAJ

Any other tips or suggestions that might be useful for this?  Cases for the badge, PCBA tips, other potential data problems?

Thanks in advance!

David

[David Dearing]

I found a few piecemeal answers to my own questions.  Replying to myself to keep it in one place:

• How long will these badges last on a single CR2032 battery?  We are looking to capture data for 12 hour "shifts."  Will these be able to handle that?
  

If this conversation from about two years ago is still relevant, it looks like a battery should last 3-4 days.  I'm not sure if that is continuously running, but seems like 12 hours should be fine.  https://groups.google.com/g/rhythm-badges/c/Tq-E8SzRXXs/m/AjoNlUoeGgAJ

• Each "shift" of our study will only have 4-5 badges actively collecting data, but those badges will be distributed over a large space with many separate rooms.  How much data can they store before needing to connect to a hub?  Will a single hub (no server) be sufficient if the badges come in contact throughout the day?
  

I'm not sure about how long the badges can accumulate data before needing to sync with a hub, but I plan to use a single hub.

 

• Our interest in these badges is primarily for capturing audio features (we have other means of tracking location data).  There will be more people talking and interacting during the study than just the 4-5 individuals that are wearing a badge.  Is it reasonable to expect the audio data to be useful?  For example, we would like to see:
  • Individual speech features for each individual (speaking, pitch, volume)
    
  • Identify conversations between those wearing badges
    
  • Conversational measures (interruptions, turn taking)  I did see this thread about turn taking, is that still current? https://groups.google.com/g/rhythm-badges/c/SPqijnJp-Ok/m/r9qIDRGICAAJ 

I'm still digging into the audio feature information.

 

[Oren Lederman]

Hi David, sorry for the late response - I'm busy with other projects now days :-)

• The badges can work for 3-4 (more like 5-6 when using two batteries) days continuously. 12 hours should be fine
• I don't remember the exact calculation but I think they are set to capture up to 1 hour of audio data before losing data. Keep in mind that downloading an hour of data can take several minutes, so if you have multiple badges coming back to the base station at the same time, you might still lose data. If you feel comfortable with embedded C you can change the storage settings to save more audio data....  but I'm personally too afraid of touching this part of the code (my students were better at this part...)
•  I'm a bit worried about the audio data you'll get. If there are other people (without badges) speaking close to the badge, or if there are other sources of sound, they might register as speaker speaking. Not sure how well this was handled in the old badges

Other than that, if you only need 10-15 badges, send me an email - I should have some. Better send it to you than having them collect dust here...

Cheers,

Oren

--
You received this message because you are subscribed to the Google Groups "Rhythm Badges" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rhythm-badge...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rhythm-badges/cfbe1171-9227-46f7-9963-dab137442732n%40googlegroups.com.