### Answer Based on the TTS provider comparison, MiniMax is your best option for spelling accuracy when dealing with numbers, emails, and links. [(1) ]() Here's the breakdown: For spelling accuracy (numbers, emails, links) : - MiniMax has the strongest spelling + most consistent tone - Cartesia offers natural sounding with stronger spelling than ElevenLabs - ElevenLabs is less reliable for exact spelling TTS Provider recommendations: - Need spelling accuracy → MiniMax (or Cartesia) - Need most natural sound → ElevenLabs (or Cartesia) - Need most consistent tone → MiniMax For STT configuration: You can configure the agent's speech-to-text mode using

stt_mode

with options like

"fast"

for speed or accuracy-focused modes. [(2) ]() [(3) ]() [(4) ]() You can also use

boosted_keywords

to bias the transcriber model for specific terms like names, brands, or technical vocabulary. Recommended combination for your use case: - TTS: MiniMax (best for spelling numbers/emails/links) - STT: Configure with

boosted_keywords

for any specific terms you need accurately transcribed - Consider adding

pronunciation_dictionary

entries for consistent pronunciation of specific words

​ To learn more, see the sources I considered: [[1] TTS provider comparison](https://docs.retellai.com/build/tts-provider-comparison) [[2] Get Voice Agent](https://docs.retellai.com/api-references/get-agent) [[3] Get Agent Versions](https://docs.retellai.com/api-references/get-agent-versions) [[4] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer \The documentation shows three STT (Speech-to-Text) modes available for Retell AI agents [(1) ]() [(2) ]() [(3) ]() [(4) ]() : -

fast

- Optimized for low latency -

accurate

- Optimized for transcription accuracy -

custom

- Allows you to configure custom STT settings For custom transcription mode: When you set

stt_mode

to

"custom"

, you must provide a

custom_stt_config

object . Based on the documentation, the custom configuration includes: -

endpointing_ms

- Controls the endpointing timing in milliseconds [(2) ]() [(3) ]() Best mode depends on your use case: - Need fastest response times? → Use

"fast"

mode [(1) ]() [(4) ]() - Need highest transcription accuracy? → Use

"accurate"

mode - Need fine-tuned control? → Use

"custom"

mode with

custom_stt_config

Additional transcription optimization options: -

vocab_specialization

- Set to

"general"

or

"medical"

for English agents [(1) ]() [(2) ]() [(3) ]() [(4) ]() -

boosted_keywords

- Bias the transcriber for specific terms like names, brands, or technical vocabulary For your use case with numbers, emails, and links, I'd recommend starting with

"accurate"

mode

combined with

boosted_keywords

to ensure the STT accurately captures those specific terms.

​ To learn more, see the sources I considered: [[1] Create Draft Agent Version](https://docs.retellai.com/api-references/create-agent-version) [[2] Get Voice Agent](https://docs.retellai.com/api-references/get-agent) [[3] Get Agent Versions](https://docs.retellai.com/api-references/get-agent-versions) [[4] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

@AI Evy but i am talking about provider for Transcription Mode

### Answer I don't have information about different STT (Speech-to-Text) provider options for Retell AI agents in the available sources. The documentation I have access to shows the

stt_mode

configuration options (

fast

,

accurate

,

custom

) and the

custom_stt_config

settings, but doesn't specify which STT provider is being used or offer provider selection options. [(1) ]() [(2) ]() [(3) ]() For the most accurate information about STT provider options and recommendations, I'd suggest: - Visit [Documentation]() - Visit [Hire certified Retell partner]() Would you like a general answer based on my knowledge instead?

​ To learn more, see the sources I considered: [[1] Create Draft Agent Version](https://docs.retellai.com/api-references/create-agent-version) [[2] Get Agent Versions](https://docs.retellai.com/api-references/get-agent-versions) [[3] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer I don't have specific information comparing Azure, Soniox, AssemblyAI, and Deepgram as STT providers for Retell AI in the available sources. The documentation I have access to covers the

stt_mode

configuration options and related settings, but doesn't include provider-specific comparisons or recommendations for those particular STT services. [(1) ]() [(2) ]() [(3) ]() For accurate guidance on which STT provider performs best for your use case (numbers, emails, links) , I'd suggest: - Visit [Documentation]() - Visit [Hire certified Retell partner]() Would you like a general answer based on my knowledge instead?

​ To learn more, see the sources I considered: [[1] Create Draft Agent Version](https://docs.retellai.com/api-references/create-agent-version) [[2] Get Voice Agent](https://docs.retellai.com/api-references/get-agent) [[3] Get Agent Versions](https://docs.retellai.com/api-references/get-agent-versions) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.