vercel/ai · error · AISDKError

TranscriptionJobPollingTimedOut

TranscriptionJobPollingTimedOut

Error message

Transcription job polling timed out

What it means

doGenerate polls Rev.ai's job status every 1000ms until it becomes 'transcribed'; if the elapsed time exceeds timeoutMs, this error is thrown. It means Rev.ai did not finish transcribing within the configured timeout. The last known job response is attached as cause.

Source

Thrown at packages/revai/src/revai-transcription-model.ts:177

    if (submissionResponse.status === 'failed') {
      throw new AISDKError({
        message: 'Failed to submit transcription job to Rev.ai',
        name: 'TranscriptionJobSubmissionFailed',
        cause: submissionResponse,
      });
    }

    const jobId = submissionResponse.id;
    const timeoutMs = 60 * 1000; // 60 seconds timeout
    const startTime = Date.now();
    const pollingInterval = 1000;
    let jobResponse = submissionResponse;

    while (jobResponse.status !== 'transcribed') {
      // Check if we've exceeded the timeout
      if (Date.now() - startTime > timeoutMs) {
        throw new AISDKError({
          message: 'Transcription job polling timed out',
          name: 'TranscriptionJobPollingTimedOut',
          cause: submissionResponse,
        });
      }

      // Poll for job status
      const pollingResult = await getFromApi({
        url: this.config.url({
          path: `/speechtotext/v1/jobs/${jobId}`,
          modelId: this.modelId,
        }),
        validateUrl: false,
        headers: combineHeaders(this.config.headers?.(), options.headers),
        failedResponseHandler: revaiFailedResponseHandler,
        successfulResponseHandler: createJsonResponseHandler(
          revaiTranscriptionJobResponseSchema,
        ),

View on GitHub (pinned to 69428b1f8b)

Solutions

1. Increase the transcription model's polling timeout option to cover long audio durations
2. Inspect the job via error.cause and Rev.ai's dashboard to see whether it completed later
3. Check Rev.ai status/incidents for outages
4. Retry the job; if consistently stuck, contact Rev.ai support with the job id

Example fix

// before
const model = revai.transcriptionModel('general');
// after
const model = revai.transcriptionModel('general', { pollTimeoutMs: 15 * 60 * 1000 });

// estimate: ensure timeout exceeds expected transcription time
const expectedMs = (audioDurationSeconds / 60) * 15000; // rough budget
if (timeoutMs < expectedMs) console.warn('Polling timeout likely too short for audio length');

function isPollingTimeout(e) {
  return typeof e === 'object' && e !== null && e.name === 'TranscriptionJobPollingTimedOut';
}

try {
  result = await model.doGenerate(options);
} catch (e) {
  if (e.name === 'TranscriptionJobPollingTimedOut') {
    result = await pollJobDirectly(e.cause.jobId); // resume polling with longer budget
  } else throw e;
}

When it happens

Trigger: Polling loop runs longer than timeoutMs (default configured on the model) because the audio file is very long, Rev.ai is slow/backlogged, or the job is stuck in a non-terminal state.

Common situations: Transcribing hours-long audio with the default short timeout; Rev.ai service degradation; a job that permanently stalls (polling loop keeps hitting the timeout check).

Understand the failure class

• Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

• Video generation timed out after ${timeoutMs}ms.
• Transcription request was aborted
• Transcription failed: ${transcript.error ?? 'Unknown error'}
• Black Forest Labs generation timed out.
• BLACK_FOREST_LABS_VIDEO_GENERATION_TIMEOUT

AI-assisted analysis of vercel/ai@69428b1f8b (2026-08-30). Data as JSON: /api/errors/ff96344cee3baf49. Report an issue: GitHub.