XEngineers/CamTalk
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request 'fix: 修复语音与文字不同步的问题，改为句子级流式播放' #68

Closed
huanghaosheng wants to merge 0 commits from develop into main
pull from: develop
merge into: XEngineers:main
Conversation 0 Commits - Files Changed - +4322 -819
huanghaosheng commented 2026-06-14 13:57:07 +08:00
Owner
Copy Link
No description provided.
huanghaosheng added 57 commits 2026-06-14 13:57:08 +08:00
fix: 修复返回给前端的 totalTokens 为0的问题并提供示例 .env b0a7ce885e
Merge pull request 'fix: 修复返回给前端的 totalTokens 为0的问题并提供示例 .env' (#49) from fix/config-file into develop 4dd89be79f 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/49
feat: 添加 Xiaomi MiMo ASR 语音识别提供者 3d3de828fc 
- 新增 MiMoService 实现 stt.Service 接口，通过 HTTP POST 调用 OpenAI 兼容的 /chat/completions 接口
- 自动将原始 PCM 数据封装为 WAV 格式（MiMo 仅支持 mp3/wav）
- 语言代码映射：zh-CN→zh、en-US→en、其他→auto
- main.go 添加 provider 选择逻辑（mimo/xiaomi → MiMo，其他 → Deepgram）
- 更新 config.yaml 使用正确的 model 名称 mimo-v2.5-asr
- 添加完整单元测试
Merge pull request 'feat: 添加 Xiaomi MiMo ASR 语音识别提供者' (#50) from fix/redis-config into develop a404fb97f9 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/50
chore: 更新 STT API Key 和 endpoint e3eb4f264b
Merge pull request 'chore: 更新 STT API Key 和 endpoint' (#51) from fix/config into develop 03a05b89a0 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/51
fix: 修复 MiMo ASR endpoint 路径重复拼接导致 404 的问题 420871cd7e 
config.yaml 中 endpoint 已包含 /chat/completions，而 mimo.go 会自动拼接该路径，
导致实际请求地址变为 .../chat/completions/chat/completions。

- config.yaml: endpoint 改为 base URL https://api.xiaomimimo.com/v1
- .env.example: STT endpoint 示例更新为 MiMo 地址
Merge pull request 'fix: 修复 MiMo ASR endpoint 路径重复拼接导致 404 的问题' (#52) from fix/stt400 into develop 4458ee82b2 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/52
fix: MiMo ASR 认证头改为 Authorization: Bearer 322061c387 
MiMo API 使用标准 Bearer Token 认证，而非自定义 api-key 头。
Merge pull request 'fix: MiMo ASR 认证头改为 Authorization: Bearer' (#53) from fix/stt400 into develop c31ecc46fc 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/53
feat: 功能优化 a7d679d04f
Merge pull request '功能优化' (#54) from fix-model into develop a6ce6d9c4f 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/54
docs: 文档与代码一致性检查与修复 6e0c67e1cb 
- 修复心跳 Bug：应用层 ping 不更新 lastPong，60 秒后连接会被错误断开
- 03-接口文档：audio/mpeg→audio/mp3、STTConfig/TTSConfig 补充 Model 字段、
  APP_ENV 环境变量名修正、配置搜索路径补充、.env 加载说明、Vite proxy 说明修正
- 02-系统架构：补充 ConfigPanel/Toast 组件、Model Router/Rate Limiter 标注规划中、
  补充 Gin 框架、MVP 存储改为 Memory、AI 服务 provider 更新、Orchestrator 伪代码对齐
- 04-技术选型：新增 AI 服务栈选型章节（STT/LLM/TTS）、PostgreSQL 标注规划中
- 06-语音交互：VAD 参数名修正、STT 改为一次性识别描述、音频编码格式补充
- 07-视觉理解：关键帧检测代码改为 TypeScript、分辨率修正、阈值逻辑统一
- 08-成本控制：变量名修正、未实现功能标注规划中、对话历史裁剪策略补充
- CLAUDE.md：同步更新技术栈、模块结构、存储策略描述
chore: 清理配置文件并移动功能创意文档 ca09a1fb72 
- .gitignore 添加 .env，防止密钥泄露
- 删除误提交的 backend/.env
- TTS 模型名修正为 mimo-v2.5-tts
- 功能创意.md 移至 docs/10-功能创意.md
docs: 编写项目 README.md 2db0e3b0b6 
- 项目简介与三层架构图
- 技术栈、项目结构树
- 快速开始指南（前端/后端启动、配置说明）
- WebSocket 协议概览与文档索引
Merge pull request 'docs: 跟进项目完成状况' (#55) from docs/update-docs into develop b8ed49be63 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/55
Reviewed-by: cfy777 <3087823110@qq.com>
feat: 添加 MiMo TTS 语音合成服务 e967d89e7e 
实现基于 Xiaomi MiMo TTS API 的语音合成 provider，使用冰糖音色。
- 新增 MiMoService 实现 tts.Service 接口
- 使用 chat/completions 格式，与 MiMo STT 保持一致
- 在 main.go 添加 TTS provider 切换逻辑（mimo/xiaomi）
- 配置文件默认音色设为冰糖
- 包含完整测试覆盖（9 个测试用例）
Merge pull request 'feat: 添加 MiMo TTS 语音合成服务' (#56) from feature/mimo-adapter into develop 5aa383cab9 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/56
feat：样式美化 70dfa1c8f8
feat: 添加 AI 服务初始化时的模型配置日志 37af0dbe06
feat: 增加前端样式优化skill和mcp工具 bf0bb83f7d
Merge pull request 'feat: 添加 AI 服务初始化时的模型配置日志' (#57) from feature/log-track into develop 1d938055e8 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/57
Merge pull request 'feat:增加前端样式优化skill和mcp工具' (#58) from frontend10 into develop b1e71be3d9 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/58
fix：修复了模型语音问题 eb7ebecfc0
Merge pull request 'fix：修复了模型语音问题' (#59) from fix-redio into develop 32f3efbb7a 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/59
fix: 修复 pipeline_test.go 中 New 函数调用参数不足的问题 00abc6c1e1
Merge pull request 'fix: 修复 pipeline_test.go 中 New 函数调用参数不足的问题' (#60) from test/pipeline-test into develop c34be3a996 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/60
feat: 扩展配置结构，新增 Session/Heartbeat/Shutdown/STT Timeout/HTTP Client/TTS Output 配置项 22aefbc216
fix: 前端 WebSocket URL 改为环境变量或自动推导，消除 localhost 硬编码 4a7cf89f29
fix: main.go 使用配置值替代硬编码 63f8cc279d 
- 版本号支持 -ldflags 构建时注入
- Session TTL/maxHistory 从配置读取
- 优雅关闭超时从配置读取
- AI 服务构造函数传入 HTTP Client 超时参数
fix: handler.go 心跳/版本号/CheckOrigin/历史上限改为配置驱动 f1ce28966c 
- 心跳间隔和超时从 config.Server 读取
- ServerVersion 从 config.App.Version 读取
- CheckOrigin 通过 config.Server.AllowedOrigins 控制
- GetHistory limit 从 config.Session.MaxHistory 读取
fix: STT 服务去除重复默认值，新增 timeout 参数 03b3566822 
- Deepgram/MiMo STT 超时从 config 传入
- 去除 model/endpoint 的 fallback 默认值，由 config 层保证
fix: LLM 服务去除重复默认值，新增 httpClientTimeout 参数 aae0629b2d 
- 去除 model/endpoint 的 fallback 默认值
- HTTP Client 超时从 config 传入
fix: TTS 服务去除重复默认值，新增 httpClientTimeout 参数 8a966c29b7 
- 去除 model/voice/endpoint 的 fallback 默认值
- HTTP Client 超时从 config 传入
fix: pipeline.go TTS 输出格式/采样率从配置读取 544716fe27 
- New() 改为接收 *config.Config 参数
- OutputFmt/SampleRate 从 config.AI.TTS 读取
test: 更新测试文件适配构造函数签名变更 96cf009228
fix: 修复新旧 TTS 语音重叠播放的问题 24a6ce5295 
- TTSPlayer 新增 stopCurrentAudio() 方法，play() 开始前先停止旧音频
- onSpeechEnd 发送新 query 前停止上一轮 TTS 播放
fix：修复语音对话配置 f6a2348c04
Merge pull request 'fix: 修复新旧 TTS 语音重叠播放的问题' (#61) from fix-redio into develop bfc896ed35 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/61
Merge pull request 'fix/hard-code' (#62) from fix/hard-code into develop 563ca12790 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/62
feat: 添加文字输入对话功能 51ed6aa563 
- 后端 WsQuery 添加 text 字段，支持文本输入模式
- Pipeline 支持文本查询时跳过 STT 直接使用输入文本
- 前端 ChatPanel 添加文字输入框，连接状态下可用
- useVisionSession 添加 sendTextMessage 方法
- 更新接口文档，添加文本输入模式说明
fix: 修复麦克风关闭后重新打开无法使用的问题 41cacaa740 
- 关闭麦克风时同时停止 VAD
- 打开麦克风时重新初始化 VAD
refactor: 前端 WebSocket 地址改为动态推导，添加 Vite 开发代理 d8630a8f26 
- websocket.ts: 移除硬编码 localhost:8080，基于 window.location 动态构建 WS URL
- vite.config.ts: 添加 /ws 和 /api 的 server.proxy 配置
- 同步更新 02-系统架构.md 和 03-接口文档.md 中的开发环境说明
Merge pull request 'refactor: 前端 WebSocket 地址改为动态推导，添加 Vite 开发代理' (#63) from fix/frontend-proxy into develop 01026d1846 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/63
fix: 修复文本输入时 LLM API 报错的问题 312e762aff 
- 移除 contentPart.Text 的 omitempty 标签
- 确保 text 字段始终存在于 JSON 请求中
Merge pull request '修复对话麦克风问题' (#64) from frontend-10 into develop dfccc824be 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/64
feat: 支持无摄像头/麦克风模式打字聊天 dffc8aa538 
- 修改 startSession 逻辑，摄像头和麦克风为可选
- 连接成功后始终显示聊天输入框
- 更新 UI 提示文字，引导用户打字对话
fix: 修复无摄像头/麦克风时聊天框不显示的问题 81c2b64e5f 
- 重构 ChatPanel 逻辑，连接后始终显示输入框
- 将欢迎消息移入消息容器内
- 确保输入框在任何状态下都可见
Merge pull request '优化无麦克风和摄像头场景的使用' (#65) from frontend-11 into develop 80e45bb8ff 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/65
feat: 添加前端 Dockerfile 和 Nginx 配置 7108d2a58b 
- 多阶段构建：node:22-alpine 构建，nginx:stable-alpine 运行
- Nginx 配置：静态文件服务 + /api /ws 反代到后端
feat: 添加后端 Dockerfile d75cdf95c8 
- 多阶段构建：golang:1.24-alpine 编译，alpine:3.20 运行
- CGO_ENABLED=0 静态链接，包含 ca-certificates 和时区数据
feat: 添加 Docker Compose 编排文件 1c74fdb894 
- frontend 和 backend 两个服务，共享 camtalk-net 网络
- backend 通过 .env 注入 API Key，不对外暴露端口
- frontend 通过 Nginx 反代对外暴露 80 端口
feat: 添加部署脚本 deploy.sh 84f5303632 
- 支持 build / up / down / restart / logs / status 命令
- 首次运行自动从 .env.example 生成 .env 模板
ci: 添加 Gitea Actions 部署工作流 65718a8460 
- PR 到 main 时验证 Docker 镜像构建
- push 到 main 时自动构建并部署，含健康检查
Merge pull request 'chore: 添加前后端 Dockerfile 和 DockerCompose 编排文件，添加部署脚本和工作流' (#66) from chore/add-docker into develop 7b3f4706e1 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/66
fix: 修复语音与文字不同步的问题，改为句子级流式播放 da5c727d8c 
之前前端 TTSPlayer 攒齐所有音频片段后才播放，导致文字全部显示后才开始语音。
改为后端每句 TTS 发送 is_last: true，前端收到每句即加入播放队列，第一句到达即开始播放。

- 后端 Chunk 结构体新增 Final 字段，区分句子结束和整轮结束
- 前端 TTSPlayer 重写为队列式播放，onended 回调自动衔接下一句
- 同步更新接口文档和测试用例
Merge pull request 'fix: 修复语音与文字不同步的问题，改为句子级流式播放' (#67) from frontend-12 into develop
Some checks failed
Backend CI / ci (pull_request) Failing after 2m14s
Details
Deploy / verify (pull_request) Failing after 1s
Details
Deploy / deploy (pull_request) Has been skipped
Details
Frontend CI / ci (pull_request) Successful in 56s
Details
d8bbdb1744 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/67
huanghaosheng added 2 commits 2026-06-14 14:05:07 +08:00
chore: 删除 deepgram_test.go 测试文件 4ebac835de
Merge pull request 'chore: 删除 deepgram_test.go 测试文件' (#69) from refactor/ai-model-refact into develop
Some checks failed
Backend CI / ci (pull_request) Successful in 1m51s
Details
Deploy / verify (pull_request) Failing after 1s
Details
Deploy / deploy (pull_request) Has been skipped
Details
Frontend CI / ci (pull_request) Successful in 1m10s
Details
1a6325698a 
Reviewed-on: http://8.161.227.145:3000/XEngineers/CamTalk/pulls/69
huanghaosheng closed this pull request 2026-06-14 14:05:39 +08:00
Some checks failed
Backend CI / ci (pull_request) Successful in 1m51s
Details
Deploy / verify (pull_request) Failing after 1s
Details
Deploy / deploy (pull_request) Has been skipped
Details
Frontend CI / ci (pull_request) Successful in 1m10s
Details

Pull request closed

Please reopen this pull request to perform a merge.
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
No items
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
cfy777 huanghaosheng
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: XEngineers/CamTalk#68
Delete Branch "develop"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?