Skip to content
GhostMind

Speech Generation

Speech Generation (Text-to-Speech / TTS) lets you convert text into audio using connected AI providers. GhostMind supports async job-based audio generation with polling, playback, and download.

Using Speech Generation

In the Playground

  1. Navigate to Playground → Speech Generation
  2. Select a provider (e.g., Google Savio)
  3. Select a voice from the available voices
  4. Optionally select a language and dialect
  5. Enter the text you want to synthesize
  6. Click Generate

Job Lifecycle

Audio generation is asynchronous:

  1. Create — Submit a generation request (returns job ID, status 202)
  2. Poll — Check job status until it’s completed
  3. Download — Fetch the audio content
created → validated → queued → processing → completed
                                             ↘ failed

Playback and Download

Once a job is completed:

  • Play — Click the play button to preview the audio
  • Download — Click download to save the audio file
  • Range support — Partial content requests are supported for streaming playback

Managing Jobs

  • Cancel — Stop a queued or processing job
  • Retry — Re-run a failed job with the same parameters
  • List — View all jobs in your workspace

Provider Details

Google Savio

Google Savio provides Arabic and multilingual TTS with:

  • Multiple voices with different dialects, genders, and age categories
  • Various languages and dialects
  • High-quality neural synthesis

See Google / Savio Speech for voice and configuration details.

Quotas

Audio generation is subject to quota limits based on your workspace’s plan. If you exceed your quota, you’ll receive a 429 error with details about your limit.

Next Steps