Compare the two embeddings

POST /v1/speaker/similarity

Returns the cosine similarity of two speaker embeddings, from -1 to 1. Your application decides how to use the score; the endpoint does not decide real-world identity.

Guide and examplesTranscription and speakers All endpoints

Request body

application/jsonRequired

embedding_anumber[]RequiredFirst embedding, exactly 256 numbers as returned by POST /v1/speaker/embeddings.

embedding_bnumber[]RequiredSecond embedding, in the same shape and from the same model as embedding_a.

modelstringModel the two embeddings came from. Defaults to EdgeSpeak/Lattice-1.Default "EdgeSpeak/Lattice-1"

Responses

200The similarity score.

objectstringspeaker_similarity

modelstring

similaritynumberCosine similarity from -1 to 1. It does not decide real-world identity.-1–1

{
  "object": "speaker_similarity",
  "model": "EdgeSpeak/Lattice-1",
  "similarity": 0.82
}

400Invalid request. Fix the request before retrying; use error.code for program logic and error.param to locate the input.

401Invalid or missing API key.

403Host or origin is not allowed, or the license was rejected. Inspect error.code to tell them apart.

503A required local model is still being prepared (model_downloading) or the service is busy (service_busy). Honor Retry-After when present.