clap-htsat-fused

Name: clap-htsat-fused
Author: laion

by laion

Contrastive Language-Audio Pretraining for audio-text retrieval

8.3Mdl/month

126likes

154Mparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

laion/clap-htsat-fused

Feature URI

mixpeek://audio_extractor@v1/laion_clap_fused_v1

Overview

CLAP learns aligned audio and text representations through contrastive learning, similar to how CLIP works for images and text. The HTSAT-fused variant uses the HTS-AT audio transformer fused with RoBERTa text embeddings.

On Mixpeek, CLAP enables semantic audio search, find audio segments matching natural language descriptions like "crowd cheering" or "rain on a roof."

Architecture

HTS-AT (Hierarchical Token-Semantic Audio Transformer) as audio encoder, RoBERTa as text encoder. Trained on AudioSet, Clotho, and other audio-text pair datasets with contrastive loss. Outputs 512-dim joint embedding space.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so clap-htsat-fused 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 vector name has to match a vector index on the collection.
          vectors: { "audio-embedding": yourVector },
          payload: { source_key: "archive/2026/asset-00412" },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// audio_fingerprint_extractor@v1 runs laion/clap-htsat-tiny
// (512-d) over a bucket, with no inference of your own.