Transcription

The transcription.realtime resource runs live transcription on a room's active session, with an optional post-session summary. (To transcribe a recording after the fact, set transcription on the recording; see Recordings.)

Start transcription

Node.js

const t = await client.transcription.realtime.start(roomId, {
  language: "en-US",
  summary: { enabled: true, prompt: "Summarize decisions and action items." },
});

Go

t, err := client.Transcription.Realtime.Start(ctx, roomID, videosdk.StartRealtimeTranscriptionParams{
	Language: "en-US",
	Summary:  &videosdk.RealtimeSummaryConfig{Enabled: videosdk.Bool(true), Prompt: videosdk.String("Summarize decisions and action items.")},
})

Options: language, modelId, webhookUrl, summary.

Stop transcription

Keyed by room id.

Node.js

await client.transcription.realtime.stop(roomId);

Go

client.Transcription.Realtime.Stop(ctx, roomID)

Get and list

Node.js

await client.transcription.realtime.get(transcriptionId);
await client.transcription.realtime.list();

Go

client.Transcription.Realtime.Get(ctx, transcriptionID)
client.Transcription.Realtime.List(ctx, videosdk.ListRealtimeTranscriptionsParams{})

list supports a status filter and pagination only, not room id.

The transcription object

id, roomId, status, start, end, and the output URLs transcriptionFilePaths and summarizedFilePaths.