smart-turn-v2

Voice Activity Detection • Pipecat • Hosted

An open source, community-driven, native audio turn detection model in 2nd version

Model Info
Batch	Yes
Real-time	Yes
Unit Pricing	$0.00034 per audio minute

Parameters

Input
Output

▶Option 1{}

object

▶Option 2{}

object

is_complete

booleanif true, end-of-turn was detected

probability

numberprobability of the end-of-turn detection

{
  "type": "object",
  "oneOf": [
    {
      "properties": {
        "audio": {
          "type": "object",
          "description": "readable stream with audio data and content-type specified for that data",
          "properties": {
            "body": {
              "type": "object"
            },
            "contentType": {
              "type": "string"
            }
          },
          "required": [
            "body",
            "contentType"
          ]
        },
        "dtype": {
          "type": "string",
          "description": "type of data PCM data that's sent to the inference server as raw array",
          "enum": [
            "uint8",
            "float32",
            "float64"
          ]
        }
      },
      "required": [
        "audio"
      ]
    },
    {
      "properties": {
        "audio": {
          "type": "string",
          "description": "base64 encoded audio data"
        },
        "dtype": {
          "type": "string",
          "description": "type of data PCM data that's sent to the inference server as raw array",
          "enum": [
            "uint8",
            "float32",
            "float64"
          ]
        }
      },
      "required": [
        "audio"
      ]
    }
  ]
}

{
  "type": "object",
  "contentType": "application/json",
  "properties": {
    "is_complete": {
      "type": "boolean",
      "description": "if true, end-of-turn was detected"
    },
    "probability": {
      "type": "number",
      "description": "probability of the end-of-turn detection"
    }
  }
}