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

# Soniox

> Configure Soniox speech-to-text in EnConvo

## Overview

Soniox is a cloud speech platform for speech-to-text, speech translation, and text-to-speech. In EnConvo, Soniox is available as a speech recognition provider for dictation and audio/video transcription.

Soniox is useful when you need high-accuracy multilingual transcription, real-time dictation, speaker diarization, and automatic language detection through your own Soniox API key.

## Supported Models

| Model                      | Mode       | Best For                                             |
| -------------------------- | ---------- | ---------------------------------------------------- |
| **Soniox STT Realtime v5** | Real-time  | Low-latency dictation and live speech input          |
| **Soniox STT Async v5**    | File-based | Audio/video file transcription and longer recordings |

<Note>
  Real-time dictation uses EnConvo's macOS realtime client with Soniox's WebSocket STT API. File transcription uses Soniox's async transcription API.
</Note>

## Setup

<Steps>
  <Step title="Create a Soniox API key">
    1. Go to the [Soniox Console](https://console.soniox.com)
    2. Sign in or create a Soniox account
    3. Open your project
    4. Create an API key
  </Step>

  <Step title="Configure credentials in EnConvo">
    1. Open **Settings** -> **Credentials**
    2. Select **Soniox**
    3. Paste your Soniox API key
    4. Validate and save the credential
  </Step>

  <Step title="Select Soniox for dictation">
    1. Open **Settings** -> **Dictation**
    2. Set the dictation provider to **Soniox**
    3. Choose **Soniox STT Realtime v5** for real-time voice input
    4. Set **Language** to **Auto** or pick a specific language
  </Step>

  <Step title="Use Soniox for file transcription">
    In the **Transcribe Audio/Video files** command, select **Soniox** and use **Soniox STT Async v5** for file-based transcription.
  </Step>
</Steps>

## Configuration

| Setting                 | Description                                              | Default                |
| ----------------------- | -------------------------------------------------------- | ---------------------- |
| **Credential Provider** | Soniox API key stored in EnConvo Credentials             | Soniox                 |
| **Model**               | Realtime or async STT model                              | Soniox STT Realtime v5 |
| **Language**            | Auto detection or a language hint                        | Auto                   |
| **Speaker Diarization** | Separate speakers when the workflow supports diarization | Available              |
| **Punctuation**         | Automatic punctuation in recognized text                 | Enabled                |

## Validate and Use

<Steps>
  <Step title="Validate the API key">
    Click **Validate** in the Soniox credential settings. If validation fails, confirm the key belongs to the correct Soniox project and has not been revoked.
  </Step>

  <Step title="Test dictation">
    For live speech input, run a short dictation test with **Soniox STT Realtime v5** before using it in longer sessions.
  </Step>

  <Step title="Test file transcription">
    For audio or video files, test a short clear recording with **Soniox STT Async v5** before processing long recordings.
  </Step>
</Steps>

## Choosing a Model

<AccordionGroup>
  <Accordion title="Use Realtime v4 for dictation">
    Choose **Soniox STT Realtime v5** when you want text to appear during live voice input. This is the best fit for Dictation, SmartBar voice input, Chat Window voice input, and short spoken commands.
  </Accordion>

  <Accordion title="Use Async v4 for files">
    Choose **Soniox STT Async v5** for recorded audio/video files, longer clips, and workflows where EnConvo transcribes after the recording is complete.
  </Accordion>

  <Accordion title="Use Auto language detection for multilingual speech">
    Set **Language** to **Auto** if you switch languages or do not know the source language. Pick a specific language when the recording is mostly one language and you want more predictable recognition.
  </Accordion>
</AccordionGroup>

## Privacy

Soniox is a cloud provider. Audio sent through the Soniox provider is processed by Soniox over the network using your Soniox account and API key.

For fully offline transcription, use a local provider such as Parakeet MLX, Qwen MLX, or Whisper MLX instead.

## Troubleshooting

<AccordionGroup>
  <Accordion title="Invalid API key">
    * Verify the key was copied from the Soniox Console without extra spaces
    * Check that the key belongs to the project you want to use
    * Regenerate the key in Soniox and update the EnConvo credential
  </Accordion>

  <Accordion title="Dictation is not real-time">
    * Confirm the selected model is **Soniox STT Realtime v5**
    * Confirm the provider is **Soniox**, not another batch transcription provider
    * Restart dictation after changing provider or model settings
    * If you still see v4 model names, update EnConvo and reopen the provider settings
  </Accordion>

  <Accordion title="Transcription language is wrong">
    * Set **Language** to a specific language instead of **Auto**
    * Check that the selected language is supported by Soniox
    * For mixed-language audio, switch back to **Auto**
  </Accordion>

  <Accordion title="No transcript from an audio file">
    * Try a common format such as MP3, WAV, FLAC, or M4A
    * Confirm the recording contains clear speech
    * Check your Soniox account usage, quota, and billing status
  </Accordion>
</AccordionGroup>

## Related

<CardGroup cols={2}>
  <Card title="Speech Recognition" icon="microphone" href="/ai/speech-recognition">
    Compare Soniox with other speech-to-text providers.
  </Card>

  <Card title="Dictation" icon="keyboard" href="/features/dictation">
    Use Soniox for voice-to-text input.
  </Card>
</CardGroup>