> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mrassistant.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Voice and providers

> Speech, models, and choosing a stack that fits your customers

MrAssistant connects **speech recognition**, **speech synthesis**, and **language models** through a provider model. Product teams care about **latency**, **accent coverage**, and **cost**; engineering teams care about **compatibility** and **rollout**.

## Speech to text

Choose transcribers that match your **languages** and **noise environments**. Retail, mobility, and support lines often have very different acoustic profiles.

## Text to speech

Pick voices that fit **brand** and **clarity**. Multilingual products may need **per-locale** voices rather than a single default.

## Language models

Balance **quality**, **speed**, and **cost**. Voice conversations reward models that behave well under **short user turns** and **interruptions**.

## Recommended defaults

When you are unsure where to start, use any **recommended configuration** your deployment exposes, then tune with real callers rather than only synthetic tests.

## Related

<CardGroup cols={2}>
  <Card title="Multilingual" icon="language" href="/voice/multilingual" />

  <Card title="Agents overview" icon="robot" href="/agents/overview" />
</CardGroup>