OpenAI Speech-To-Text service for the Freeswitch.
Features: vad, flexible adjustments.
Available through: asr_interface, commnad api
If you have troubles with building this module, pay attention to this issue and the comment, maybe you faced the same thing.
<extension name="openai-asr"> <condition field="destination_number" expression="^(3222)$"> <action application="answer"/> <action application="play_and_detect_speech" data="/tmp/test2.wav detect:openai"/> <action application="sleep" data="1000"/> <action application="log" data="CRIT SPEECH_RESULT=${detect_speech_result}"/> <action application="hangup"/> </condition> </extension>session.ttsEngine= 'openai'; // requires: mod_openai_tts session.asrEngine= 'openai'; var txt = session.sayAndDetectSpeech('Hello, how can I help you?', 10); consoleLog('info', "TEXT: " + txt);freeswitch> openai_asr_transcript /tmp/test.[wav|mp3] [key=altKey mode=altModel] +OK: How old is the Brooklyn Bridge