awesome-convAI

The repository contains a curated list of Conversational AI modules such as Automatic speech recognition (ASR), Dialog systems (Chatbots) and Speech synthesizer (TTS)

Automatic Speech recognition(ASR) system or Speech-to-text(STT)

Various ASR engines are available as open-source. Based on the underlying architecture, they are classified as Hidden Markov Models (HMM) and Deep learning models.

HMM ASR:

• Cmusphinx
• Kaldi

DL ASR:

• DeepSpeech
• Coqui STT
• SpeechBrain
• ESPnet

Commercial ASR (Cloud based solutions):

• Dragon Speech Recognition
• Azure Speech
• Google STT

Speech synthesizer or Text-to-Speech(TTS)

TTS helps to synthesize audio-speech from text. Eventhough lots of previous TTS systems sound robotic, recent developments in TTS lead to human-like speeches.

• Coqui TTS
• wavenet
• Tacotron
• Deepvoice
• voiceloop

Conversational agent (NLU)

• Google Dialogflow
• Amazon Lex
  • Amazon lex is restricted to English language.
• IBM Watson
• NVIDIA NeMo
• Wit
• Microsoft LUIS
• Rasa
  • Rasa does not have ASR support directly, but connecting with an external ASR engine is possible.
• Xenioo