Is decoupling speech synthesis from synchronization units worth building?
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 35/100
- Issue 类型
- 功能
- 描述清晰度
- 基本清楚
- 活跃度
- 活跃
调研方向
Read the linked questions in the Map issue, starting with the answers on continuity, cost and latency, and placement accuracy; then compare language coverage and the size of the synchronization-unit change. No files or tests are named. Done means recording a reasoned go or no-go, with the required design for a follow-up map or spec—or the provider, price or model change that would reopen a no-go.
由索引模型根据 Issue 内容生成。
描述
Question
Is decoupling speech synthesis from synchronization units worth building, now that overlap and cut is withdrawn (ADR 0028)? It is judged against speech policy 6, and it should reach the continuity overlap and cut showed by ear: flow at 9 of 10 boundaries, against 4 of 10 for policy 6 (#291).
Decide go or no-go, with reasons, from the map's answers:
- continuity and chunk size: How long can one Gemini 3.8 TTS request get before a fresh start inside it?;
- cost and latency: What would larger narration chunks cost against speech policy 6, and at what latency?;
- placement accuracy: How accurately does speech recognition place synchronization units inside a larger narration chunk?;
- language coverage: Which speech recognition models with word timing cover Cup's narration languages?, How can synchronization units be placed in audio without speech recognition? and Which narration languages have Cup's conversions had?, including whether the intersection of the Gemini TTS languages and one speech recognition model's languages is good enough, and what the rest get;
- the size of the change: What does a synchronization unit become when it no longer has its own audio segment?.
A "go" names what the follow-up map or spec must design; a "no-go" names what would have to change (a provider, a price, a model) to reopen it.
Part of Map: decouple speech synthesis from synchronization units.
- 主要语言
- TypeScript
- 星标
- 1
- 派生
- 0
- PR 合并指标
- 30 天内没有已合并 PR
环境准备
这个项目没有提供开发容器、Dockerfile 或贡献指南,环境需要你自己搭建:先看它的 README,通用步骤见我们的新手贡献指南。
从这里开始
- 先读完整个 Issue,再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库,在一个分支上完成修改。
- 提交 Pull Request,并在描述里引用这个 Issue 编号。
patricktree/cup 的其他 Issue
-
wayfinder:grilling
难度 2/5 1-3 小时 新手友好度 72/100
patricktree/cup#484 ·
-
wayfinder:grilling
难度 2/5 1-3 小时 新手友好度 48/100
patricktree/cup#487 ·
-
wayfinder:grilling
难度 5/5 一周以上 新手友好度 25/100
patricktree/cup#486 ·
-
wayfinder:grilling
难度 5/5 一周以上 新手友好度 25/100
patricktree/cup#485 ·
-
wayfinder:grilling
难度 5/5 一周以上 新手友好度 25/100
patricktree/cup#483 ·
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 78/100
vercel-labs/ui#865 ·
维护者通常 1 天内回复
-
bug
难度 2/5 1-3 小时 新手友好度 63/100
outline/outline#14638 · 1 条评论 ·
维护者通常 1 天内回复
-
bot:ai-assisted component:platform-js status:untriaged
难度 2/5 1-3 小时 新手友好度 78/100
midnightntwrk/platform-js#69 ·
维护者通常 1 天内回复
-
bug
难度 2/5 1-3 小时 新手友好度 67/100
维护者通常 1 天内回复