Prepare text for speech
Use this action when you are about to speak text that was written to be read, and want it to sound like something a person would actually say. It returns text only — it produces no audio.
In the editor
In the action picker it is under AI and voice, as "Rewrite text the way people speak". In the flow, a step reads like: Rewrite Step 1 · choices · 0 · message · content to be read aloud.
Under What it does:
- Title of this step (shown in logs), in the step's header: optional.
- Text to rewrite: the written text, as Text or Expression. Insert data adds values from the flow, usually an earlier step's answer. 8,000 characters at most: a longer text, or an empty one, makes the step fail.
A note under the text says what the step does: an AI model rewrites text made to be read (headings, lists, tables, emoji) the way a person would say it, and the step gives back text, not audio. To hear it, pass this step's text to Turn text into speech.
Later steps can read:
text: The rewritten text, ready to pass to "Turn text into speech".model: The AI model that rewrote the text.- Succeeded: Yes when this step did not fail.
Every step also has This step runs, If this step fails, the run, Edit as fields, View as JSON and Done: see The step editor.
Earlier name: Prepare spoken text.
Best for
- Preparing an agent or LLM reply before speaking it
- Speaking a template or a rich-text field that contains formatting
- Any step whose text is headed for Turn text into speech
This is not the same as removing the formatting. A table has no spoken form, so stripping the markup does not give you speakable text — it gives you word soup, reading the cells out in a row or dropping the answer altogether. This action asks a small, inexpensive model to say the same thing the way a person would, while your original text stays untouched for the transcript.
Use something else when
| You want | Use |
|---|---|
| The audio itself | Turn text into speech — run this action first, then that one |
| New content, not a rewrite of text you already have | Ask an AI model — it takes your own prompt and model |
Main fields
The step's own fields, behind Edit as fields and in the API and MCP:
| Field | What it does |
|---|---|
| Markdown | The written text to rewrite for speech, at most 8000 characters. Usually the reply from an earlier step, such as workflow.step(0).choices[0].message.content |
| Purpose | Optional label for what this step is for, shown in the logs |
What later steps can use
| Value | What it is |
|---|---|
workflow.step(N).text | The spoken form. Feed it straight into Turn text into speech's Text field |
workflow.step(N).model | The model that produced the rewrite |
Limits
| Limit | Value |
|---|---|
| Input length | 8000 characters |
| Speech requests | 60 per minute, per company |
The rewrite has an output budget of 1200 tokens, which does not guarantee it fits the 4096-character synthesis limit. Check size(workflow.step(N).text) before synthesis; split oversized text into chunks of at most 4096 characters.
The 60-per-minute ceiling is shared with voice mode in the app and with Turn text into speech, so a prepare-then-speak pair costs two of those requests per run.
Cost
This step counts toward your plan's AI token usage, not your audio usage. It runs on a small, inexpensive model chosen for you — you cannot pick the model, because the job is a mechanical rephrase rather than reasoning, and it should not cost reasoning prices.
When it fails
| Code | What happened |
|---|---|
spoken_text_too_long | The input is over 8000 characters. Shorten it or split it across steps |
Voice mode in the app quietly falls back to plain text when this rewrite fails, but a workflow has no such fallback: a failure here fails the step. If you want the same graceful degradation, gate the synthesis step on workflow.stepOk(N) and pass the original text through on the other branch.