Generate and play speech
Use Text to speech to hear your own text or Listen to a completed transcript. Speech is off by default and has its own connection, model, and voice. You can use it without a microphone or transcription setup.
For local speech, add MagpieTTS to NeMo. For another service, save a speech connection first.
Set up and speak
Section titled “Set up and speak”- Open Text to speech → Configure speech. The sidebar’s Connection or Model row also opens Speech options.
- Choose a connection, model, model profile, and voice. Use Refresh models and supported Refresh voices to read metadata, or enter exact IDs.
- Turn on Enable text to speech and choose Save.
- Enter your text, then choose Speak or press Ctrl + Enter. Enter alone adds a new line. Audio plays when generation succeeds.
Preview a voice before saving
Section titled “Preview a voice before saving”In Speech options, turn on Enable text to speech and choose Preview beside the voice selector. This sends a fixed sample phrase to the selected service using your unsaved choices. It does not save settings or enable playback elsewhere. Stop and preview again to hear changes; save to apply them to the composer and Listen actions.
Control playback
Section titled “Control playback”| Action | What happens |
|---|---|
| Pause / Resume | Pauses or continues existing audio |
| Playback slider | Moves when you release it; use arrows, Home, or End with the slider focused |
| Restart in Playback actions | Replays existing audio from the beginning |
| Stop | Cancels active generation or stops playback |
| Save generated speech | Exports the full generated audio, regardless of playback position |
| Clear completed audio in Playback actions | Removes the retained audio from memory |
Seeking while playing continues playback. Seeking while paused or after completion leaves audio paused until Resume. It uses audio already in memory and sends no new generation request.
You can edit or clear the text during generation or playback; those edits do not change the current audio. After generation finishes, Speak generates the current draft and replaces the previous audio, even if playback is paused. Playback and Save continue to use the existing audio until then.
Listen to a transcript
Section titled “Listen to a transcript”Choose Listen on a completed Voice or Audio-file result, even with history off, or on a retained history entry. This sends the chosen text to your speech endpoint. Playback starts only when you ask; Freehand does not automatically read each transcript or provide a voice conversation mode.
If speech fails
Section titled “If speech fails”Open Details beside playback controls for the explanation and Speech settings. Check the system output device and volume. In the composer, Try again uses the current draft; failure does not erase it.
If Restart rewinds but cannot open the output device, audio stays paused at the beginning. Correct the device and use Resume or Restart. You can still save or clear retained audio. A failed rewind leaves the existing session and position in place.