#We are getting a TimedOut response from ElevenLanbs api when using scribe_v2. Here is the error:h

1 messages · Page 1 of 1 (latest)

real robinBOT
#
AI-generated Answer for tomac1.

Short answer

  • Your ReadTimeout is most consistent with a long-running, synchronous Scribe v2 batch transcription exceeding your HTTP client’s wait window. ElevenLabs notes that Speech‑to‑Text (batch) processes proportionally to audio length (typically 20–50× real‑time), so multi‑minute files can take minutes to complete, which can outlast default client timeouts. (elevenlabs.io)
  • There is no current outage: the ElevenLabs status page reports “We’re fully operational” as of today (March 26, 2026). Recent ASR/STT incidents earlier in March were resolved. (status.elevenlabs.io)

What to do

  • Use asynchronous STT: call Create transcript with webhook=true so the request returns immediately and the finished transcript is delivered to your configured webhook(s). This is explicitly recommended for long‑running jobs. (elevenlabs.io)
  • If you need live/low‑latency results, switch to Scribe v2 Realtime over WebSocket. (elevenlabs.io)
  • For large inputs, you can also pass cloud_storage_url instead of uploading the file in the request. (elevenlabs.io)
  • Reference limits and behavior: STT supports large files and long durations (batch), and the platform also supports webhook delivery for asynchronous completion. (elevenlabs.io)

Sources

Sources:
ElevenAPI - ElevenLabs the most powerful AI audio APIs
ElevenLabs Status
Create transcript | ElevenLabs Documentation
Realtime | ElevenLabs Documentation
Create transcript | ElevenLabs Documentation
Transcription | ElevenLabs Documentation
ElevenLabs Status
ElevenAPI - ElevenLabs the most powerful AI audio APIs
Create transcript | ElevenLabs Documentation
Asynchronous Speech to Text | ElevenLabs Documentation
Realtime | ElevenLabs Documentation

frail vapor
#

I will try to increase limit in Elevenlabs SDK

real robinBOT
#
AI-generated Answer for tomac1.

“Increasing limit” in the ElevenLabs SDK won’t resolve a ReadTimeout for scribe_v2. The SDK/docs don’t expose a “limit” parameter related to Scribe v2 HTTP timeouts. Use one of the documented, timeout-safe patterns instead:

  • Make the batch STT call asynchronous: set webhook=true (and optionally webhook_id). The request returns immediately and the transcript is delivered via webhook. (elevenlabs.io)
  • Or submit the job and poll for the result with Get transcript (/v1/speech-to-text/transcripts/{transcription_id}). (elevenlabs.io)
  • If you need continuous results without long HTTP waits, use Scribe v2 Realtime over WebSocket (wss://api.elevenlabs.io/v1/speech-to-text/realtime). (elevenlabs.io)
  • Plan client wait/poll intervals based on documented STT throughput (typically 20–50× real time). Large files will take proportionally longer to process. (elevenlabs.io)

If you want, I can outline the minimal code changes for either the webhook or realtime approach.

Sources:
Create transcript | ElevenLabs Documentation
Get transcript | ElevenLabs Documentation
Realtime | ElevenLabs Documentation
ElevenAPI - ElevenLabs the most powerful AI audio APIs

frail vapor
#

Wrong answer.
Increasing the timeout limit from the default 240 to 300 in Elevenlabs finally helped solve the problem 😉

real robinBOT
#
AI-generated Answer for tomac1.

Thanks for the update—glad you’re unblocked.

For future robustness with Scribe v2 batch transcription:

  • Expect long-running jobs: Speech-to-Text processing time scales with audio length (20–50× real-time) and batch STT supports large inputs (up to 10 hours, 3 GB), so synchronous calls can run for minutes. (elevenlabs.io)
  • Prefer the asynchronous pattern to avoid client read time limits: enable webhooks on batch STT (e.g., pass webhook=true to speech_to_text.convert) and handle the completion POST instead of waiting on the HTTP response. (elevenlabs.io)
  • If you need low-latency transcription, use Scribe v2 Realtime over the WebSocket endpoint (wss://api.elevenlabs.io/v1/speech-to-text/realtime). (elevenlabs.io)

Sources:
ElevenAPI - ElevenLabs the most powerful AI audio APIs
Asynchronous Speech to Text | ElevenLabs Documentation
Realtime | ElevenLabs Documentation

frail vapor
#

We will switch to async sooner or later. We are fine now 😉