Speech Recognition

SgneepBox jeOS integrates all functions to make life comfortable for those unfamiliar with technologies related to voice recognition. Offline recognition is entrusted to the powerful STT CMUSphinx engine. SgneepBox deals with all NLU, NLP, and microphone optimization operations (Respeaker4Mic compatible), ACG, Noise Suppression, VAD.


STT and TTS in Speech Recognition

Insights:

Compatible Mic Array : Respeaker4 SeeedStudio

Speech Recognition System : CMUSphinx


SgneepBox Script command


Set STT Language

speech_recognition *lang

*language, possible value : en-us | it-IT

Make Personal Language Model (LM):

lm_clear

lm : *your_phrase

lm_process


Make Personal Speech To Text (STT) model :

stt_clear

voice_word *id *speech_word

voice_type *id *[subscript | speak | meteo | musicCTRL | videoCTRL]

voice_subscript *id *subscriptname

voice_stopmusic *id

voice_response *id *ttsID


Make Personal Text To Speech (TTS) model:

tts_phrase *id *lang : *response_phrase


Start Microphone Array (SeeedStudio Respeaker4)

sst_script_name *scriptFileName

sst_respeaker_close

sst_respeaker_run

Example:

#.with SgneepBox Editor insert this code in subscript
#
#
speech_recognition en-us
# connect word 'car' with subscript *scriptFileName
voice_word w1 the car
voice_type w1 subscript
voice_subscript w1 scriptFileName
#--------------------------------
#
# Speech To Text
# set STT stop and halt word recognited from microphone array
voice_word w2 stop
voice_type w2 stopmusic
voice_word w3 halt
# close mp3 music player voice_type w3 stopmusic
#--------------------------------
#
# Set TTS phrase
tts_phrase ph1 en : hello my name is jenny
# set STT word for call TTS phrase voice_word w4 hello world
voice_type w4 speak
voice_response w4 ph1
#--------------------------------
# start Respeaker4
sst_script_name DEMO_TTS_RESPEAKER
stt_respeaker_close
stt_respeaker_run
#--------------------------------
Powered by Sgneep.com