@mcut/transcription-assemblyai

@mcut/transcription-assemblyai adapts the AssemblyAI client to the mcut TranscriptionProvider interface.

Install

bun add @mcut/transcription-assemblyai

Types

AssemblyAITranscriptLike

interface AssemblyAITranscriptLike {
  text?: string | null
  words?:
    | Array<{
        text: string
        start: number
        end: number
        confidence?: number
        speaker?: string | null
      }>
    | null
  utterances?:
    | Array<{ text: string; start: number; end: number; speaker?: string | null }>
    | null
  language_code?: string | null
  audio_duration?: number | null
}

The fields of an AssemblyAI transcript that normalizeAssemblyAIResult reads, with timestamps in milliseconds.

AssemblyAIProviderOptions

interface AssemblyAIProviderOptions {
  apiKey?: string
  client?: AssemblyAI
  speakerLabels?: boolean
  params?: Partial<Omit<TranscribeParams, 'audio'>>
  id?: string
}

Options for createAssemblyAIProvider, an API key or configured client, speaker labeling, extra request params, and a provider id.

Functions

normalizeAssemblyAIResult

function normalizeAssemblyAIResult(transcript: AssemblyAITranscriptLike): TranscriptResult

Converts an AssemblyAI transcript to a TranscriptResult, mapping words with confidence and speaker to words and utterances to segments.

createAssemblyAIProvider

function createAssemblyAIProvider(
  options: AssemblyAIProviderOptions = {},
): TranscriptionProvider

Returns a TranscriptionProvider that uploads audio to AssemblyAI with speaker labels on by default, and throws when neither an API key nor a client is available.