VR app for Pico Neo 3 headset for the Listening Effort project.
Earlier versions of this project were built for the HTC Vive. The last compatible version for HTC Vive can be found on the htc-vive branch.
- Extract the APK file from the release ZIP.
- Plug the Pico into a Windows PC with a USB cable
- Enable USB access on the Pico. This only needs to be done once:
With the headset on select the 'Pico Button' to bring up the dock and select the 'Quick settings' icon in the bottom right corner. Select the 'Settings' icon, which is located in the top right corner and navigate to the 'Developer' button. Turn on 'USB Debug'. Click on 'USB Connection' and select 'File transfer'. - Copy the APK file onto the Pico - you can just put it in the root of the “Internal Storage” directory that shows up in the Pico (look in “This PC” in Windows Explorer)
- On the Pico, open File Manager and go to APKs, and you can install it from there.
On a Mac, you may need to use Android File Transfer.
- Install Unity 2021.3.26f1.
- In Build Settings, set platform to Android.
- Plug in Pico, enable developer permissions if needed, then select Pico as Run Device in Build Settings.
- Build and run.
When the app starts it will display a screen showing the folders where you can copy your own videos, scripts, HRTFs and impulse responses (BRIR files). The log files will be recorded to an adjacent folder.
The app will copy over a demo script to this location which you can use as a template for your own scripts.
The first thing the app does is validate all user videos to ensure there are no problems midway through an experiment. The controls will be greyed out until it completes this.
Scripts may be defined in YAML and selected in the app. Audio will be played back on headset using the 3DTI Toolkit. The app will then automatically cycle through all specified videos and invite the participant to record their response using the headset microphone. Full session data is logged.
In OSC mode, the app is fully controlled by a Max patch over an OSC connection. Audio is not played, scripts are not read and data is not logged.
An example Max patch demonstrating this can be found in Utilities/Unity controller.maxpat. (On the htc-vive branch this may be in a different location).
- When the app is opened, it will check all video files to ensure they can be opened. This takes a long time, and it may be necessary to cache these findings and simply confirm the filename and timestamp haven't changed on previously validated videos.
- While videos are being checked, it remains possible to select a script. If the script references videos that haven't yet been checked, then the script will fail validation with an error stating that these videos cannot be found. To avoid this, wait until the video check is complete before selecting a script. (If you've already selected the script, then change to a different script then change back to the intended script to retrigger validation.)
The app in the master branch currently asks participants to record their audio response but does not perform any transcription on this.
Research was conducted on the feasibility of automatic transcription of the recorded speech. The full findings can be read in 2024-02-15 Report on feasibility of speech recognition.pdf.
The prototypes mentioned in the report can be found as follows:
Within the speech-recognition branch.
A python implementation using Whispercpp.py can be found in the master branch under the Utilities/LogPostprocessor directory.