Summary

Speech to Text by ElevenLabs provides highly accurate automatic speech recognition models, including Scribe v2 and Scribe v2 Realtime, enabling transcription of live and recorded audio in over 90 languages. The service offers industry-leading accuracy with low latency, supporting diverse accents and recording conditions.

Features

• Scribe v2 Realtime: Real-time transcription with sub-150 ms latency, ideal for live conversations, meetings, and AI applications across 90+ languages.
• Scribe v2: Converts speech from audio and video files into editable text, captions, and subtitles with high accuracy.
• Advanced capabilities: Voice activity detection, speaker and entity detection, dynamic audio tagging, keyterm prompting, and sensitive information redaction.
• Multilingual support: Supports over 90 languages with varying accuracy tiers, from excellent to moderate word error rate (WER).
• API and SDK access: Integration of speech-to-text functionalities into products via fully streaming APIs and SDKs.
• Security and compliance: Enterprise-grade data protection with encryption, and supports SOC 2, HIPAA, GDPR, EU data residency, and zero retention modes.
• Flexible deployment: Cloud and on-premise options, with granular team permissions and support for custom workflows.
• Pricing: Multiple subscription tiers from free to pro with monthly billing, pricing per transcribed hour, and enterprise plans.

Use Cases

Suitable for transcription of podcasts, videos, interviews, meetings, call centers, AI agent interactions, captioning for social media, accessibility tools, and any scenario requiring live or recorded speech transcription.

Additional Information

The technology relies on deep neural networks trained on large multilingual datasets, delivering accuracy that outperforms other speech-to-text solutions such as Whisper and Deepgram in benchmarks.

Features like speaker diarization enable distinguishing multiple speakers automatically, and real-time transcription is available via dedicated streaming architecture.

The service includes tools for editing and managing transcripts, captions, and subtitles, facilitating content repurposing and accessibility compliance.