Hi, this is Sonetho. ⚡
Quick recap of the January 13 ElevenLabs webinar. ElevenLabs is no longer just "an AI voice service" — they're positioning as an all-in-one AI creative platform.
From Studio 3.0 (with top-tier video models like Sora 2 and Veo 3) to Scribe v2 (more accurate than the human ear) — deep-dive on everything they announced.
1. Studio 3.0 — All-in-One Creation
Headline drop: Studio 3.0. Core idea: workflow consolidation.
🎥 Studio 3.0's 3 big innovations
- Top-tier video models integrated: Google Veo 3, OpenAI Sora 2, Kling, Ideogram — the best video/image generators are now inside ElevenLabs Studio. No separate subscriptions needed.
- One-stop timeline: enter text → [TTS + SFX + BGM + captions + video] all auto-generate on one timeline.
- Inline editing: don't like one section? Drag and edit just that part, no full re-generation.
This isn't an incremental feature — it's the output of strategic partnerships with Disney, NVIDIA, and Adobe.
2. Scribe v2 — #1 STT Accuracy Globally
Scribe v2 stomps the existing STT field. The announced WER (Word Error Rate) data:
| Model | WER | Note |
|---|---|---|
| ElevenLabs Scribe v2 | 2.2% | #1 globally |
| GPT-4o Transcribe | 2.7% | - |
| Gemini 1.5 Pro | 3.0% | - |
| Deepgram Nova 3 | 6.9% | - |
* Lower = better (averaged across English/French/Spanish)
Scribe v2 killer features:
- Audio Event Tagging: tag laughter, applause, footsteps as text events
- Smart Diarization: identifies speakers even on overlapping speech
- Word-level Timestamps: per-word timing for perfect caption sync
3. Enterprise security & scaling
🔒 Security & Compliance
- SOC 2 / ISO 27001 — top-tier infosec certs
- Zero Retention — optional, no data persists server-side
- GDPR-compliant — full EU data protection alignment
🤝 Collaboration
- Project sharing + approval workflows across teams
- Granular access control for internal teams and agencies
4. [Q&A] Live questions from the session
Q. When does V3 ship?
A. Final stages. Late January, at latest mid-February.
Q. Can we control breath / pitch after generation?
A. Yes — heavily requested. Fine-tune parameters for post-generation tweaking are in research, update coming.
Q. Localized UI roadmap?
A. Yes — multi-language UI landing this year. No more juggling translators.
Bottom line: imagination → reality, faster
Webinar's core message: "Just imagine. AI handles production." One text prompt now generates video + voice + sound simultaneously.
If you want to ride this wave first, try Studio 3.0 now.
Try Studio 3.0 — 50% off first month
▶ Start now📚 Related
❓ Frequently Asked Questions (FAQ)
Q1. What changes with Studio 3.0?
A. The core focus is on workflow integration.
Video and image generation models such as Google Veo 3, OpenAI Sora 2, Kling, and Ideogram are now integrated into the studio.
Simply by entering text, voice, sound effects, background music, subtitles, and video are automatically generated on a single timeline.
You can edit specific sections by dragging them, without needing to regenerate the entire project.
Q2. What is the announced accuracy of Scribe v2?
A. The Word Error Rate (WER) revealed in the webinar shows ElevenLabs Scribe v2 at 2.2%, ranking first.
GPT-4o Transcribe followed at 2.7%, Gemini 1.5 Pro at 3.0%, and Deepgram Nova 3 at 6.9%.
Lower figures indicate higher accuracy, based on the average of major languages including English, French, and Spanish.
Q3. What are the unique features of Scribe v2?
A. Three main features were introduced.
Audio Event Tagging, which converts sounds like laughter, applause, and footsteps into text tags.
Smart Diarization, which identifies individual speakers even when they overlap.
Word-level Timestamp, which ensures precise subtitle synchronization by capturing the timing of every individual word.
Q4. How is the security for enterprise use?
A. The company announced it has achieved SOC 2 and ISO 27001 certifications.
A Zero Retention option is available for enterprises that prefer not to have data stored on servers.
It is also fully GDPR compliant.
The platform supports team project sharing, approval workflows, and granular access control management.
Q5. What was said regarding the v3 release date and Korean language support?
A. During the webinar Q&A, it was stated that v3 is in the final stages and will be available by late January or February at the latest.
They mentioned that Korean UI support is being prepared for later this year.
They also noted that they are researching parameters to allow for fine-tuning of features like breath control and pitch adjustment after generation.
Sonetho. ⚡