Qwen3-ASR-0.6B-hf

Name: Qwen3-ASR-0.6B-hf
Author: Qwen

by Qwen

782M multilingual speech recognition, small enough to run over a whole archive

782Mparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

Qwen/Qwen3-ASR-0.6B-hf

Feature URI

Overview

Speech recognition is the step that makes audio and video searchable at all. Nothing downstream works until the words exist as text, and the cost of that step decides whether you transcribe everything or only the files someone asks about.

At 782M parameters this is a small model by current standards, which is the interesting property. Whisper-large-class models are more accurate and considerably more expensive per hour of audio; a model this size changes the arithmetic on transcribing an entire back catalogue rather than a sample of it.

The card declares Chinese, English, Cantonese, Arabic, German, French, Spanish and Portuguese, so it is genuinely multilingual rather than English with extras.

Architecture

Qwen3ASRForConditionalGeneration, model type qwen3_asr, 782,426,112 parameters. Automatic-speech-recognition pipeline. Declared languages: zh, en, yue, ar, de, fr, es, pt.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so Qwen3-ASR-0.6B-hf runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns_your_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection_id: "col_your_collection",
      documents: [
        {
          document_id: "asset-00412",
          // The model produces text, so it lands in payload. Give the
          // collection a text vector index and embed that text to make it
          // searchable rather than only filterable.
          payload: { extracted_text: modelOutput, source_key: "archive/2026/asset-00412" },
          vectors: { "text-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// universal_extractor@v1 runs google/gemini-embedding-2
// (3072-d) over a bucket, with no inference of your own.