博客
来自团队的产品动态、使用指南与工程笔记。
Synthetic Voice Disclosure: A Practical Publishing Checklist
Decide when and how to disclose synthetic narration, document consent and provenance, follow platform controls, and avoid misleading voice use.
Accessible Audio Content: A Practical Captions and Transcript Guide
Plan captions, transcripts, visual descriptions, playback controls, and audio QA from the start so podcasts, videos, and narration reach more people.
GPT-5.6 Sol '50% Off' Is a Gateway Promo, Not an OpenAI List-Price Cut
A Hacker News headline said GPT-5.6 Sol was cut 50%. OpenAI's list price did not move. Here is what actually changed on gateways, and what did not.
Speko Launches as an OpenRouter for Voice AI. What Changes for Voice Stacks
Speko, a YC S26 company, routes STT, LLM, and TTS from published language-specific benchmarks. Here is what that layer is — and is not — for people following GPT-Live-1.
How to Write E-Learning Narration That Helps People Learn
Design e-learning narration around learner actions, clear explanations, useful pauses, accessibility, and a review workflow—not slide reading.
Podcast Intro Script Templates That Sound Like Your Show
Write a concise podcast intro with a clear promise, host identity, episode context, music cues, and adaptable templates for solo and interview shows.
How to Write a YouTube Voiceover Script That Matches the Edit
Plan a YouTube voiceover around visuals, timing, captions, and retakes with a scene-by-scene script workflow that survives the final edit.
Text-to-Speech Pronunciation Guide: Fix Names, Acronyms, and Numbers
A practical TTS pronunciation workflow for names, acronyms, homographs, dates, and technical terms—without breaking the rest of your script.
Voiceover Script Word Count: Estimate Any Video or Audio Length
Estimate voiceover duration from word count, choose a realistic speaking pace, and plan pauses, retakes, and visual timing without guesswork.
How to Format a Text-to-Speech Script for Natural Narration
Turn ordinary copy into a clean TTS script with better pauses, pronunciation, sentence rhythm, and a repeatable listening-based editing workflow.
GPT-Live-1 是什么？OpenAI 全双工语音模型详解
GPT-Live-1 是 OpenAI 新出的能边听边说的语音模型。全双工是什么意思、和 Advanced Voice Mode 有何区别、谁能用——一篇讲清。
朴伊桑