> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bland.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Speech Studio

> Write, direct, and synthesize speech in the dashboard with the Norm director, performance tags, and a live editor.

Speech Studio produces Bland Speech audio without code. You describe what you want the voice to say and how it should sound. The studio sends each complete script through the same [`POST /v2/tts`](/api-v2/post/tts) endpoint you can call directly.

Open it at [app.bland.ai/dashboard/tts/studio](https://app.bland.ai/dashboard/tts/studio).

## What's in the studio

<CardGroup cols={2}>
  <Card title="Script editor" icon="pen-to-square">
    A live transcript editor with per-turn voice assignment, inline performance tags, and multi-take generation.
  </Card>

  <Card title="Norm director" icon="wand-magic-sparkles">
    A chat panel that writes and edits your script for you. Describe the delivery in plain language and Norm updates the editor in real time.
  </Card>

  <Card title="Voice picker" icon="microphone">
    Pin featured voices, search your cloned voices, filter by version, and preview any voice with a reactive orb.
  </Card>

  <Card title="Session history" icon="clock-rotate-left">
    Every session is saved with its script and past takes. Reload the page and you land back in the session you were in.
  </Card>
</CardGroup>

***

## Write a script

The editor is a single transcript field. Type what you want spoken, and press **Generate** (or `Cmd`/`Ctrl` + `Enter`) to synthesize.

Pick a voice from the top **Voice** picker before generating. Every new turn defaults to a featured voice so you can generate immediately without configuring anything.

### Performance tags

Wrap performance cues in square brackets to shape how a line is delivered:

```text theme={null}
[warm] Hey, good to hear from you.
[sighs] I know. [pause] Let me look at that for you.
```

Type `[` in the editor to open an autocomplete list of supported tags. Use arrow keys and `Enter` to insert one. Tags behave as a single atomic block: one backspace deletes the whole tag, and arrow keys step over it.

You can also open the **Effects** panel to browse tags and click any one to drop it in at the cursor.

### Multiple turns

Add turns for a back-and-forth read. Each turn gets its own voice picker, so you can alternate speakers within a single script. Unassigned turns fill in with a stable default voice so you can keep writing.

### Takes

Choose **1x**, **2x**, or **3x** in the transport bar to generate that many takes from the same script in one run. Each take gets its own waveform; select a take to scope playback, timestamp, and download to just that one.

***

## Direct with Norm

Norm is the studio's built-in director. Open the side panel and use the **Chat** tab to describe what you want, and Norm writes or edits the script in the editor while it streams.

Norm is a single-speaker director. It works on the currently selected voice from the top voice picker.

Use Norm to:

* Draft a script from a brief ("30-second warm voicemail asking them to call back")
* Rewrite a line to hit a specific tone ("make the middle beat slower and a little tired")
* Add or remove performance tags across the script
* Iterate turn-by-turn without leaving the editor

Norm streams updates live, so you'll see the transcript rewrite as it works.

<Tip>
  Norm and the editor stay in sync. Anything you type by hand carries through the next Norm turn, and anything Norm writes you can edit directly.
</Tip>

***

## Sessions and history

Every studio session is persisted locally with its script, voice assignments, and finished takes. Switch to the **History** tab in the side panel to jump between sessions or start a new one with the **+** button in the header.

* Sessions are titled from the first line of the script until you rename one.
* Reloading the page returns you to the session you were last in.
* Past takes replay from cached audio; sessions whose audio was evicted show a **Transcript-only** tag and can be regenerated on demand.

***

## Pricing

Speech Studio bills against pay-as-you-go TTS credits. See the current per-character rate and refill options under **Credits** in the TTS sidebar, or in [Billing & Plans](/platform/billing).

***

## Do the same thing from the API

Everything in the studio is powered by the public TTS API. To reproduce a studio generation programmatically:

```bash theme={null}
curl -X POST https://api.bland.ai/v2/tts \
  -H "authorization: YOUR_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "text": "[warm] Hey, good to hear from you. [pause] Let me pull that up.",
    "voice": "29158307-9893-4149-8a75-bc9ce313d64e",
    "audio": { "container": "wav" }
  }' \
  --output line.wav
```

See [Synthesize Speech (HTTP)](/api-v2/post/tts) for the full request shape, and [List Voices](/api-v1/get/voices) to find a voice ID. For token-level input and conversational interruptions, use [Realtime Speech (WebSocket)](/api-v2/post/tts-ws).