OpenAI 发的 GPT-Transcribe 在音频转录上比 Whisper 错误率低一半，多语言场景提升明显。
GPT-Transcribe 在 Context Aware ASR 基准上，语义准确率从无自由格式上下文时的 41.6% 提升至 45.2%。在 Common Voice 的 22 种语言上，GPT-Transcribe 转录错误率为 19.27%，而 Whisper（whisper-1）为 40.37%。在 Real-World Audio Recording 基准的 9 种语言上，GPT-Transcribe 错误率为 8.98%，Whisper 为 15.21%。
GPT-Transcribe also improved on the Context Aware ASR benchmark, with semantic accuracy increasing f...
GPT-Transcribe also improved on the Context Aware ASR benchmark, with semantic accuracy increasing from 41.6% without free-form context to 45.2% with it. Across 22 languages on Common Voice, GPT-Transcribe achieved a 19.27% transcription error rate, compared with 40.37% for Whisper (whisper-1). Across nine languages on the Real-World Audio Recording benchmark, it achieved an 8.98% transcription error rate, compared with 15.21% for Whisper (whisper-1). 💬 2 🔄 1 ❤️ 12 👀 2075 📊 3 ⚡