#Tool calling capabilities degraded when using third-party TTS providers

1 messages · Page 1 of 1 (latest)

quasi gull
#

If I use ElevenLabs or OpenAI as TTS providers, the tool calling isn't nearly as functional if I use the same agent but with deepgrams TTS

terse escarpBOT
#

Hey there! This channel allows you to ask Kapa.ai for assistance. But it looks like you haven't connected your Discord profile to your Deepgram account yet. You can do so at https://community.deepgram.com - by doing so, you'll be able to ask questions in #1115960287183511643 if you need human assistance!

native bladeBOT
#

Hi @quasi gull. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

Based on the available knowledge sources, there is no explicit documentation or discussion that directly addresses a degradation of tool (function) calling capabilities when using third-party TTS providers (such as ElevenLabs or OpenAI) with Deepgram Voice Agent, compared to using Deepgram’s own TTS.

The documentation does confirm that Deepgram Voice Agent supports multiple TTS providers, including Deepgram, ElevenLabs, and OpenAI, and provides configuration examples for each. The function calling (tool calling) capabilities are described as part of the agent’s LLM configuration, and are not documented as being dependent on the TTS provider. The TTS provider is configured separately from the LLM and function calling logic, and there is no mention of reduced functionality or limitations in tool calling when using third-party TTS providers in the official documentation or example configurations TTS Models, Function Calling.

If you are experiencing degraded tool calling functionality in practice, it may be due to implementation details, integration issues, or specific limitations not covered in the public documentation. The knowledge sources do not provide enough information to confirm or explain such a degradation.

If this is a critical issue, I recommend reaching out to Deepgram support or providing detailed feedback through their official channels, as there may be undocumented limitations or recent changes affecting this behavior.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: