Voice Synthesis

The Voice Synthesis page turns any text into spoken audio using your configured TTS provider. Use it to generate listening material, model pronunciation, or build shadow-reading practice clips.

The Workspace

The page is organized into three columns:

1. Input — Type or paste any text. Optionally preload a translation result via the translationKey URL parameter.
2. Settings — Choose:
   • Language — The language of the input text.
   • Voice — A speaker from the provider's voice catalog. Default voice is selected automatically based on language and provider.
3. Result — The generated audio, with play controls and history.

Generating Audio

1. Type or paste text into the input column.
2. Pick a language and voice.
3. Click Generate. The result appears with an embedded player and a download link.

The first run for a given (text, voice, language) is cached locally for fast replay.

Voices and Providers

Voices depend on the active TTS provider in Settings → AI Services. Common providers include Microsoft Azure Speech, OpenAI TTS, and Cloudflare Workers AI.

If a provider is unavailable, the voice selector will be empty — switch to a configured provider in Settings.

History

Every TTS result is added to your audio history. The history panel shows:

• A truncated preview of the input text
• The voice and language used
• A timestamp
• A replay button

Click any entry to load it back into the result column.

Linking from Smart Translation

If you start with a translation, you can send the translation result directly to Voice Synthesis. The translationKey URL parameter preloads the result text, and the target language is selected automatically.

Tips for Best Results

• Match language to text — Selecting the wrong language produces poor pronunciation and weird pauses.
• Use punctuation — Periods, commas, and question marks influence the prosody.
• Shorten very long inputs — TTS quality degrades for very long passages; chunk into paragraphs.