# Insert Silent Pauses with <break>

## 1. Why insert breaks?

Even great prosody sometimes needs explicit timing. The `<break>` tag adds a **precise, silent pause** in speech without changing words or phonemes. Use it to improve clarity (e.g., after numbers or acronyms), add dramatic effect, or separate clauses where punctuation alone isn’t enough.

---

## 2. Tag syntax

| Element/Attr | Required | Description                                |
| ------------ | -------- | ------------------------------------------ |
| `<break/>`   | ✓        | Self-closing tag that inserts a pause.     |
| `time`       | ✓        | Duration of the pause, e.g., `300ms`, `1s` |

**Example**
`Your code is 9 4 2 1 <break time='400ms'/>Please repeat it back.`

**Notes**

* Self-closing form: `<break time='…'/>`
* Place `<break/>` **between** tokens; do not put it **inside** `<phonemes>…</phonemes>`.

---

## 3. Duration format & guidance

* **Units supported:** milliseconds (`ms`) and seconds (`s`)
  Examples: `150ms`, `250ms`, `0.5s`, `1s`
* **Natural range:** `80ms`–`1200ms` is usually best
  (Very small values may be rounded; very large values may be clamped.)

**Heuristics**

* 80–150 ms → subtle micro-pause (comma-like)
* 200–400 ms → short phrase break
* 500–800 ms → sentence-level pause
* 1.0–1.5 s → dramatic/section break

---

## 4. Request format

```jsonc
POST /text_to_speech/streaming HTTP/1.1
Host: api.async.com
Content-Type: application/json
X-Api-Key: <YOUR_API_KEY>

{
  "model_id": "async_flash_v1.0",
  "transcript": "Welcome to Async.<break time='400ms'/>Let's get started.",
  "voice": { "mode": "id", "id": "e0f39dc4-f691-4e78-bba5-5c636692cc04" },
  "output_format": {
    "container": "raw",
    "encoding": "pcm_s16le",
    "sample_rate": 44100
  }
}
```

### Curl shortcut

```bash
curl -L https://api.async.com/text_to_speech/streaming \
  -H "Content-Type: application/json" \
  -H "X-Api-Key: $ASYNC_API_KEY" \
  -d '{
        "model_id":"async_flash_v1.0",
        "transcript":"Welcome to Async.<break time='\''400ms'\''/>Let'\''s get started.",
        "voice":{"mode":"id","id":"e0f39dc4-f691-4e78-bba5-5c636692cc04"},
        "output_format":{"container":"raw","encoding":"pcm_s16le","sample_rate":44100}
      }' | ffplay -f s16le -ar 44100 -
```

---

## 5. Interaction with other tags

* **With `<digits>`**
  OTPs/phone numbers benefit from a pause after the block:
  `Your OTP is <digits>6 1 4 9 2</digits><break time='300ms'/>It expires soon.`

* **With `<phonemes>`**
  Use breaks **around** a phoneme block, not inside it:
  `Welcome to <phonemes>ˈeɪ.sɪŋk</phonemes>.<break time='350ms'/>Enjoy the demo.`


---

## 6. Best practices

1. **Prefer punctuation first**; add `<break>` when punctuation doesn’t yield the timing you want.
2. **Use sparingly** to avoid choppy delivery.
3. **Tune by ear**—start at `250–400ms` for phrase breaks and adjust.
4. **Keep rhythm consistent** across similar sections (e.g., list items).
5. **Accessibility**—short pauses after numeric blocks/acronyms improve intelligibility.

---

## 7. Examples

**Sentence-level pause**

```text
This is Async.<break time='600ms'/>A better way to build voice.
```

**List pacing**

```text
You will hear three tones:<break time='250ms'/>low,<break time='250ms'/>mid,<break time='400ms'/>and high.
```

**Digits + pause**

```text
Confirm code:<digits>3 8 1 7</digits><break time='300ms'/>Resending now.
```

**Brand name with phonemes + pause**

```text
Welcome to <phonemes>ˈeɪ.sɪŋk</phonemes>.<break time='350ms'/>Let’s begin.
```

---

## 8. FAQ

**Q:** Is `time` the only attribute?
**A:** Yes. Use `time` with `ms` or `s`. (Attributes like “strength” are not used in Async markup.)

**Q:** Can I chain multiple `<break>` tags?
**A:** You can, but prefer a single `<break>` with the total duration.


**Q:** Any hard limits?
**A:** Extremely small values may be rounded; very large values may be clamped. For naturalness, keep to `80–1200ms` unless you intentionally need longer silence.