Hi community 👋
A new feature released today caught my attention: custom ASR dictionaries in bot flows (V3 engine) 👏
This is a great step forward, especially for handling:
• Technical terms
• Acronyms
• Brand names
• Industry-specific vocabulary
Being able to guide the recognition layer like this can significantly improve bot accuracy and overall customer experience.
According to the documentation, this capability is currently tied to the V3 engine:
https://help.genesys.cloud/articles/speech-to-text-stt-engines-overview/
And this is particularly relevant for voice scenarios, where the challenge is not interpretation, but ensuring that what the customer says is transcribed correctly in the first place.
For digital channels, the text is already structured, but in voice, transcription accuracy is everything.
However, this also brings an important discussion.
As we know, Genesys currently provides multiple STT engines:
• V1 (Google)
• V2 (Microsoft)
• V3 (AWS)
And while V3 is the most recent, in real-world voice scenarios, especially in Brazil, V1 and V2 can sometimes perform better, particularly due to:
• Regional accents
• Slang
• Linguistic variations
So here's my question:
👉 Are there any plans to extend custom dictionary support to V1 and V2 engines as well?
Because in many cases, the best-performing engine is not necessarily V3, and having dictionary support across all engines would allow much more flexibility and accuracy in bot design.
Would love to hear how others are approaching this today 🚀
#ConversationalAI(Bots,VirtualAgent,etc.)#DigitalChannels------------------------------
Mateus Nunes
Tech Leader Of CX at Solve4ME
Brazil
------------------------------