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

# StepAudio 2.5 ASR

StepAudio 2.5 ASR is a 4B-parameter speech recognition model. It introduces Multi-Token Prediction (MTP) technology to predict multiple tokens per step in parallel, maintaining SOTA transcription accuracy while dramatically reducing serial wait cycles: a 5-minute audio clip can be fully transcribed in 1 second.

<Columns cols={3}>
  <Card title="Online Demo" icon="play" href="https://stepaudiollm.github.io/step-audio-2.5-asr/">
    Visit the official demo page to experience the model's capabilities firsthand.
  </Card>

  <Card title="API Quick Start" icon="rocket" href="#quick-start">
    Minimal runnable curl example.
  </Card>
</Columns>

## Key Information

<Columns cols={3}>
  <Card title="Architecture">
    4B MTP
  </Card>

  <Card title="Engine-side RTF">
    ≈ 0.0053<br />\~19 seconds to transcribe 1 hour of audio
  </Card>

  <Card title="API Pricing">
    See pricing details
  </Card>
</Columns>

## Core Capabilities

<Columns cols={2}>
  <Card title="Low-latency Inference">
    Introduces MTP (Multi-Token Prediction) technology. Predicting multiple tokens per step in parallel boosts throughput by 400% and cuts latency by 60% compared with traditional ASR. A 5-minute audio clip is fully transcribed in 1 second.
  </Card>

  <Card title="SOTA Transcription Accuracy">
    Deeply optimized on 4B parameters. Achieves industry-leading Chinese and English error rates across diverse scenarios including news, meetings, and noisy environments.
  </Card>
</Columns>

## Use Cases

Voice Agents, large-scale transcription services, real-time captions / live streaming.

## API Endpoint

StepAudio 2.5 ASR offers two access methods overseas: one-shot HTTP + SSE (model `stepaudio-2.5-asr`) and real-time bidirectional WebSocket streaming (model `stepaudio-2.5-asr-stream`).

<Card title="Speech Recognition (Streaming Output)" icon="file-lines" href="/docs/en/api-reference/audio/asr-sse">
  `POST /v1/audio/asr/sse`<br />Submit audio as Base64 once; the server streams transcription back over SSE. Supports PCM / OGG / MP3 / WAV, Chinese and English recognition, and the `enable_itn` parameter.
</Card>

<Card title="Streaming Speech Recognition (Bidirectional Streaming)" icon="file-lines" href="/docs/en/api-reference/audio/asr-stream">
  `WSS /v1/realtime/asr/stream`<br />Real-time bidirectional WebSocket streaming, suitable for conversations and live captioning. Supports server-side VAD and per-word timing.
</Card>

## Pricing

Step Plan subscribers can use this model directly. See [pricing & rate limits](/docs/en/guides/pricing/details) for full details.

## Quick Start

```bash theme={null}
curl https://api.stepfun.ai/v1/audio/asr/sse \
  -H "Authorization: Bearer $STEP_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: text/event-stream" \
  -d '{
    "audio": {
      "data": "base64_encoded_audio",
      "input": {
        "transcription": {
          "model": "stepaudio-2.5-asr",
          "language": "zh",
          "enable_itn": true
        },
        "format": {
          "type": "pcm",
          "codec": "pcm_s16le",
          "rate": 16000,
          "bits": 16,
          "channel": 1
        }
      }
    }
  }'
```

The server emits `transcript.text.delta` events incrementally and ends with `transcript.text.done`.

## Related Resources

<Columns cols={2}>
  <Card title="Demo Page" icon="flask" href="https://stepaudiollm.github.io/step-audio-2.5-asr/">
    Product demo page.
  </Card>

  <Card title="Model Card" icon="circle-info" href="https://stepaudiollm.github.io/step-audio-2.5-asr/model-card/">
    Model card with architecture and benchmark details.
  </Card>

  <Card title="Speech Recognition (Streaming Output) API" icon="file-lines" href="/docs/en/api-reference/audio/asr-sse">
    Full parameters, response events, and error handling.
  </Card>
</Columns>