gpt-4o-transcribe at a glance
gpt-4o-transcribe and gpt-4o-mini-transcribe improved on Whisper's word error rate, and gpt-4o-mini-tts let developers describe how a voice should speak.
What changed with gpt-4o-transcribe
On March 20, 2025, OpenAI added new audio models to its API for building voice agents. gpt-4o-transcribe and gpt-4o-mini-transcribe are speech-to-text models that OpenAI said beat its original Whisper models on word error rate across several benchmarks, with better handling of accents, noise and fast speech.
The new gpt-4o-mini-tts text-to-speech model is steerable: for the first time developers could instruct not only what to say but how to say it, such as sounding like a calm, sympathetic support agent. All three models build on GPT-4o and GPT-4o mini and were opened to all developers.
Why gpt-4o-transcribe matters for businesses
For businesses running phone or chat assistants, more accurate transcription and a controllable voice tone improve call summaries and customer experience, though disclosures and consent rules for recorded or AI-voiced calls still apply.
Put it to work: AI Voice Agents
Sources for this story
- Introducing next-generation audio models in the API (opens in a new tab)Official
OpenAI · openai.com
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #speech-to-text
- #text-to-speech
- #openai
- #voice-agents
