Sound Localization trials
Concerning speech recognition and audio detection test, the work is in progress.
The current test solutions are :
- Concerning the sound localization, an ad hoc microphone system based on 8 MEMS MP34DT01,
2 shields X-NUCLEO-CCA02M1 and 2 STM32F401RE have been designed and is used with hark sound framework.
- For the high quality sound recording we used a ORTF solution based on 2 Shure PGA81 microphone an ART Dual Pre USB for analogic to digital conversion and a Rode Stereo Bar SB20 for the ORTF positionning.
- For the Speech to text, different solutions are under comparison including, sphinx, julius and the vocon sdk from Nuance.
- Concerning the text analysis for questions/answers phases and the command analysis, we develop solutions based on nltk python library.
- The choice of the Speech to Text is not define at this time.