Production TestedSpeech-to-Text

ElevenLabs Scribe v2

ElevenLabs' realtime STT offering — poor quality and slow for Arabic.

Not Recommended

Poor quality and poor latency for Arabic. Not recommended for any Arabic STT use case.

ElevenLabs Scribe v2 Realtime was tested for Arabic speech recognition via LiveKit Inference. Despite being marketed for real-time use, it showed both poor quality and high latency for Arabic, performing similarly to Google Chirp 3 on speed but with significantly worse transcription accuracy.

Benchmarks

Latency

Avg EOU Delay2000ms–2500ms
Best Case2000ms
Worst Case2500ms

Quality

RatingPoor
Arabic Dialect Support
Saudi Arabic

Described as 'shit quality' in production testing. Not viable for Arabic.

Features

Real-time streaming transcription
Multiple language support
LiveKit inference integration
StreamingLiveKit Plugin

PricingFree Tier Available

PlanPriceUnit
Starter$5per month

Integration

SDKs
PythonNode.js
API Style

WebSocket streaming

Documentation

Verdict

Poor quality and poor latency for Arabic. Not recommended for any Arabic STT use case.

Best For

Pros

  • LiveKit plugin available
  • Part of ElevenLabs ecosystem (TTS bundle)

Cons

  • Poor Arabic transcription quality
  • High latency (2-2.5s EOU)
  • No advantage over better alternatives
Visit ElevenLabs Scribe v2

Go to https://elevenlabs.io

Compare with other Speech-to-Text providers