Skip to main content
Updated Nov 16, 2026

Turn text into speech

Use this action when you have text and want a spoken audio file of it, saved in your company storage.

In the editor​

In the action picker it is under AI and voice, as "Make an audio file from text". In the flow, a step reads like: Read Step 3 · text aloud, as MP3, or, with a voice picked, Read Step 3 · text aloud in the voice Support voice, as MP3.

Under What it does:

  1. Title of this step (shown in logs), in the step's header: optional.
  2. Text to say: as Text or Expression. Insert data adds values from the flow, such as an earlier step's text. 4,096 characters at most, about five minutes of speech: a longer text, or an empty one, makes the step fail.
  3. Voice: a voice profile, picked from a list. Left empty, the company's default voice profile is used, and AutoTalk's own voice when there is none.
  4. Format: MP3 audio or Opus audio (OGG).

The audio is ready when this step ends: later steps get it as a file, which a message step can send.

Later steps can read:

  • file: The audio file, for a message step or "Inspect media".
  • bucket: The bucket the audio file is in.
  • fullPath: The audio file's path in the bucket.
  • size: The size of the audio file, in bytes.
  • contentType: The content type: audio/mpeg for MP3, audio/ogg for Opus.
  • format: The format that was made: mp3 or opus.
  • cache: "hit" when audio made earlier was reused (no provider was called and no speech time was counted), otherwise "miss". The file is saved either way.
  • Counted on (billedMeter): What the speech was counted on: tts_ms (AutoTalk's provider, billed), audio_ms_byok (your own provider, not billed), or empty when earlier audio was reused. Read it before relying on the billed milliseconds.
  • billedMs: The milliseconds counted; 0 when earlier audio was reused. They are billed only when the meter is tts_ms: with your own provider this is an estimate and is not billed.
  • 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: Synthesize speech.

Best for​

  • Speaking an order confirmation, an appointment reminder or an IVR prompt
  • Producing a voice note to send back over a channel
  • Turning text an earlier step produced into an audio file you can attach or store

Use something else when​

You wantUse
A transcript of speech, not speech from textTranscribe audio — it goes the other direction
To re-encode audio you already haveConvert media — this action only makes new audio from text
To speak text that came from an LLM or a rich-text fieldPrepare text for speech first, then this action

Main fields​

The step's own fields, behind Edit as fields and in the API and MCP:

FieldWhat it does
TextThe text to speak, at most 4096 characters. Usually workflow.step(N).text from an earlier step
Voice profileWhich voice to use. Leave it empty to use your company default
Formatmp3 (the default) or opus
PurposeOptional label for what this step is for, shown in the logs

Only Text takes a CEL expression. Voice profile, Format and Purpose are literal fields. Omit the voice profile to use the company default and omit Format to use mp3. A malformed profile identifier or an unsupported format is rejected; CEL typed into those fields is not evaluated. Select valid values in their controls.

The audio is ready when the step finishes​

Unlike Transcribe audio and Convert media, this action does not queue a job. It synthesizes the audio inline, so there is nothing to poll and nothing to wait for.

By the time the next step runs, workflow.step(N).file names an audio object that is really there. You can attach it, store it, or pass it straight to another media step in the same run.

What later steps can use​

ValueWhat it is
workflow.step(N).fileThe {bucket, fullPath} reference every media action produces. Feed it to Send a message to a contact attachments, a file field, or Inspect media
workflow.step(N).bucketThe storage bucket the audio was written to
workflow.step(N).fullPathThe full path of the audio object
workflow.step(N).sizeThe size of the audio in bytes
workflow.step(N).contentTypeaudio/mpeg for mp3, audio/ogg for opus
workflow.step(N).formatThe format that was actually used
workflow.step(N).cachehit when a recent identical request was reused, otherwise miss
workflow.step(N).billedMetertts_ms when the synthesis was billed, audio_ms_byok when it ran on your own provider key and was not billed, or empty on a cache hit
workflow.step(N).billedMsThe milliseconds recorded on billedMeter, and 0 on a cache hit

Read the outcome with workflow.step(N).billedMs, not workflow.step(N).status — status is reserved for HTTP status codes, and this action deliberately never writes it. Use workflow.stepOk(N) to test whether the step succeeded.

Voices​

A voice profile holds the voice, provider and settings used to speak your text. Profiles live in the Speech Studio's Profiles tab, where the company owner creates and edits them, and one of them can be marked as the company default.

This action picks a voice in this order:

  1. The Voice profile you named on the step
  2. Your company's default profile
  3. The built-in platform voice, so synthesis works before you configure anything

A voice pinned to a channel is not used here. That pin only applies inside the app, where a conversation identifies its channel. A workflow has no conversation to resolve, so if you want a specific voice, name the profile on the step.

Limits​

LimitValue
Text length4096 characters, roughly 700 words or five minutes of speech
Speech requests60 per minute, per company

Longer text is rejected, not split. There is no server-side chunking and no joining of audio, so text over the limit fails rather than producing a partial file. If you need more, split the text across several steps and accept one audio file per step.

The 60-per-minute ceiling is shared with voice mode in the app and with Prepare text for speech, so a prepare-then-speak pair costs two of those requests per run. A tight synthesis loop can use up the allowance and interrupt someone using voice in the app at the same time — pace the workflow if it synthesizes in bulk.

Cost​

Synthesis counts toward your plan's audio usage, based on the length of the text rather than the length of the finished audio. On the default OpenAI voice, whose provider price follows the audio, a speed below 1 counts for the longer audio it produces. Voices differ a lot in price: some cost several times more per minute than others, so the voice profile you pick changes the bill. On AutoTalk's key, synthesis costs at least 2× the provider's published price.

Profiles that use your own provider key are tracked but not billed by AutoTalk — your provider bills you directly. Check workflow.step(N).billedMeter before reading workflow.step(N).billedMs: on your own key that number is an estimate for your own records, not a charge.

Speaking the same text again shortly afterwards is usually free. A repeat of identical text with the same voice and format within about ten minutes reuses the earlier audio, makes no provider call, and reports cache: hit with billedMs: 0. This is a best-effort reuse, not a guarantee — a busy period or a restart can turn it back into a normal, billed call. Audio produced on your own provider key is never reused this way.

A cache hit still writes a new file. Every call stores its own audio object, so a workflow that speaks the same text in a loop keeps growing your storage usage even while the synthesis itself costs nothing.

When it fails​

CodeWhat happened
speech_missing_textThe Text expression resolved to nothing
speech_text_too_longThe text is over 4096 characters. Split it across steps
speech_format_invalidFormat is something other than mp3 or opus
voice_profile_not_foundThe profile named on the step does not exist
voice_profile_id_invalidVoice profile does not hold a valid profile reference — often an expression that was not evaluated
limits_reachedYour plan's audio allowance for the month is used up
rate_limit_errorMore than 60 speech requests in the same minute. Worth retrying shortly
storage_quota_exceededThe audio was produced but your storage is full, so it could not be saved
speech_synthesis_failedThe voice provider could not be reached or refused the request. Worth retrying

storage_quota_exceeded is the one to watch: the audio was synthesized and counted before the save was attempted, so the usage is spent even though no file arrived. Free up storage before retrying.

Elsewhere​

The same capability is available outside workflows as the synthesize_speech MCP tool and as POST /v1/speech on the API. All three produce the same audio and the same fields; only where the file is stored differs.