# Speaches Speech Server API Exposure Detection Module

id: speaches-api-exposure
info:
  name: Speaches Speech Server API Exposure
  author: sif
  severity: medium
  description: Detects an exposed Speaches speech server (the successor to faster-whisper-server) reachable without authentication; api keys are disabled by default, so its model list leaks the loaded speech-to-text and text-to-speech models and the open api accepts unauthenticated transcription and synthesis
  tags: [speaches, faster-whisper, whisper, speech, stt, tts, ai, inference, api, exposure, recon]

type: http

http:
  method: GET
  paths:
    - "{{BaseURL}}/v1/models"

  matchers:
    - type: regex
      part: body
      regex:
        - '"object"\s*:\s*"list"'

    - type: word
      part: body
      words:
        - "\"owned_by\""

    - type: regex
      part: body
      regex:
        - '"task"\s*:\s*"(automatic-speech-recognition|text-to-speech|speaker-embedding|voice-activity-detection|speaker-diarization)"'

    - type: status
      status:
        - 200

  extractors:
    - type: regex
      name: speaches_model
      part: body
      regex:
        - '"id"\s*:\s*"([^"]+)"'
      group: 1