phonevis
Introduction
Offers a visualisation of speaker phonetics (loudness, frequency, pitch) for the last spoken words at a time in an effort to show what else may be possible in the furture. Offers GUI to choose speakers, toggle word filter and maximum words to display and whether mean or median of the speaker statistics is used for filtering. Saves the settings of each iteration during live session in the MongoDB collection data_phonevis. A replay of live session and recreation of singular plots based on this collection is possible.
Visualization
Main Contributor
Alke Kaul
Technical Framework
Python 3.8 implementing:
Requirements
- Speech2Text container must run and output transcriptions.
- phonentics container must run and output collection “data_phonetics” in MongoDB
- CHROME is recommended, since the Speech2Text container currently does not run in Firefox.
- for debugging and information messages in console the phonevis_dockerfile.yml must run the app “unbuffered” (-u tag in call of dockerfile)
Settings
- LANGUAGE=EN/DE choosen system language: Phonevis currently only supports English and German.
- S2T_STREAMING_DEBUG=false must be set to false so S2T will generate nessecary output
- Applied Sentiment Dictionaries: