> For clean Markdown of any page, append `.md` to the page URL.
> For a complete documentation index, see https://docs.sarvam.ai/llms.txt.
> For full documentation content in one file, see https://docs.sarvam.ai/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.sarvam.ai/_mcp/server.

# How to set buffer size to start processing in Streaming TTS with `min_buffer_size`

> Define when the TTS engine should start processing text from the buffer.

The `min_buffer_size` parameter sets the **minimum number of characters** that must accumulate in the buffer before the TTS engine begins converting text into audio.
We buffer incoming text until it reaches this threshold before processing and chunking begins.

> **Note**
>
> This parameter is available only in the **WebSocket Streaming API**. It is not supported in the REST API.

This helps balance latency with natural sentence completion during real-time TTS streaming.

### **Parameter Details**

* **Type:** Integer
* **Range:** `30` to `200`
* **Default:** `50`
* **Purpose:** Minimum character length that triggers buffer flushing and TTS processing.

---

### How It Works

* When the buffer reaches `min_buffer_size`, the text is **automatically processed** and streamed as audio.
* If the buffer does **not** reach the threshold, the text is **held** until:
  * More characters arrive, **or**
  * A **flush command** is sent.

### Manual Flush Option

A **flush command** forces the TTS engine to **immediately process** the current buffer, even if it hasn’t reached the `min_buffer_size`.

---

### Practical Example

Suppose `min_buffer_size = 50`, and you send an 80-character sentence in two parts:

| Input Chunk | Characters | Result                |
| ----------- | ---------- | --------------------- |
| First part  | 60         | Processed immediately |
| Second part | 20         | Held in buffer        |

To process the remaining 20 characters, send a **flush message**.

---

### **Example Streaming API code**

```python
import asyncio
import base64
from sarvamai import AsyncSarvamAI, AudioOutput
import websockets

async def tts_stream():
    client = AsyncSarvamAI(api_subscription_key="YOUR_SARVAM_API_KEY")

    async with client.text_to_speech_streaming.connect(model="bulbul:v3") as ws:
        await ws.configure(
            language_code="hi-IN", 
            speaker="shubh",
            min_buffer_size= 80
        )
        print("Sent configuration")

        text = (
            "भारत की संस्कृति विश्व की सबसे प्राचीन और समृद्ध संस्कृतियों में से एक है।"
            "यह विविधता, सहिष्णुता और परंपराओं का अद्भुत संगम है, "
            "जिसमें विभिन्न धर्म, भाषाएं, त्योहार, संगीत, नृत्य, वास्तुकला और जीवनशैली शामिल हैं।"
        )


        await ws.convert(text)
        print("Sent text message")

        await ws.flush()
        print("Flushed buffer")

        chunk_count = 0
        with open("output.mp3", "wb") as f:
            async for message in ws:
                if isinstance(message, AudioOutput):
                    chunk_count += 1
                    audio_chunk = base64.b64decode(message.data.audio)
                    f.write(audio_chunk)
                    f.flush()

        print(f"All {chunk_count} chunks saved to output.mp3")
        print("Audio generation complete")

        
        if hasattr(ws, "_websocket") and not ws._websocket.closed:
            await ws._websocket.close()
            print("WebSocket connection closed.")


if __name__ == "__main__":
    asyncio.run(tts_stream())

# --- Notebook/Colab usage ---
# await tts_stream()

```

> Define when the TTS engine should start processing text from the buffer.