v0.1.8

Fix

• Fixed: language selection always reverted to Japanese→Chinese after saving — now correctly persists user's chosen language pair

v0.1.7

What's New

API Key Authentication

• Switched from X-Api-App-Key + X-Api-Access-Key to the new single X-Api-Key header

Corpus Support (Hot Words, Glossary, Replacements)

• Hot Words: Guide the model to prioritize recognizing specific words (jargon, names, brands)
• Glossary: Define fixed translation pairs for consistent terminology
• Replacement Words: Fix common misrecognitions in source and translated subtitles

Voice Cloning

• New TTS mode selector: Off / Preset Voice / Clone Speaker Voice
• Clone mode replicates the input speaker's voice for TTS output

Separate Settings Window

• Settings now open in an independent window (not constrained by main window size)
• Organized into two tabs: General and Dictionary
• Each corpus field includes help text explaining its purpose

Other Improvements

• Speaker change detection (spk_chg) support
• Token usage billing (UsageResponse) parsing
• UpdateConfig (event 201) encoder for mid-session corpus updates
• 8-language i18n for all new UI strings

Downloads

Platform	File
macOS (Apple Silicon)	TransEcho_0.1.7_aarch64.dmg
Windows (x64, installer)	TransEcho_0.1.7_x64-setup.exe
Windows (x64, MSI)	TransEcho_0.1.7_x64_en-US.msi

v0.1.6

Changes

• Auto-pause on sustained silence — after 60s of silence, automatically disconnects the API to stop token consumption. Reconnects automatically when speech resumes (~1-2s delay)
• Unified session lifecycle — merged audio pipeline and event loop into a single task supporting disconnect/reconnect
• Improved echo suppression — timestamp-based cooldown (150ms) replaces boolean flag, covering WASAPI device buffer latency
• Fixed platform string — "macos" correctly mapped to "macOS" for Volcengine API compatibility
• Enhanced deduplication — concatenation matching detects API-accumulated repeated text
• Performance — eliminated per-frame Vec allocation in audio pipeline hot path

Token Savings

Scenario	Before	After
2h meeting, 1h speech + 1h silence	45,000 tokens	22,875 tokens (-50%)
Left running overnight (8h idle)	180,000 tokens	375 tokens (-99.8%)

Downloads

Platform	File
macOS (Apple Silicon)	TransEcho_0.1.6_aarch64.dmg
Windows (x64, MSI)	TransEcho_0.1.6_x64_en-US.msi
Windows (x64, EXE)	TransEcho_0.1.6_x64-setup.exe

v0.1.5

Changes

• Improved TTS echo suppression — replaced AtomicBool with timestamp-based cooldown (150ms), covering WASAPI device buffer latency so residual TTS audio is fully suppressed
• Silence keepalive — sustained silence now sends a small burst of zero frames every ~5s to keep WebSocket alive, instead of continuously flooding the API
• Fixed platform string — std::env::consts::OS returns "macos" (lowercase); now correctly mapped to "macOS" for Volcengine API compatibility
• Enhanced deduplication — added concatenation matching to detect API-accumulated repeated text
• Performance — eliminated per-frame Vec allocation in audio pipeline hot path

Downloads

Platform	File
macOS (Apple Silicon)	TransEcho_0.1.5_aarch64.dmg
Windows (x64, MSI)	TransEcho_0.1.5_x64_en-US.msi
Windows (x64, EXE)	TransEcho_0.1.5_x64-setup.exe

v0.1.4

Changes

• Fix: Windows 对话空档期不再重复最后一句话（提高静音检测阈值 + 迟滞机制）
• Feature: 设置界面显示版本号
• Signing: macOS 版使用 Developer ID 证书签名并通过 Apple 公证，可直接打开

Downloads

• macOS (Apple Silicon): TransEcho_0.1.4_aarch64.dmg
• Windows (x64): TransEcho_0.1.4_x64-setup.exe / TransEcho_0.1.4_x64_en-US.msi

TransEcho v0.1.3

TransEcho v0.1.3

Bug 修复

• 修复对话空档期重复最后一句话的问题：新增 RMS 静音检测，当系统音频处于静音状态时不再向 API 发送空白帧，避免 ASR 引擎反复识别残留语音上下文。此修复同时适用于 macOS 和 Windows。

下载

平台	文件
macOS (Apple Silicon)	TransEcho_0.1.3_aarch64.dmg
Windows (x64)	TransEcho_0.1.3_x64-setup.exe (推荐)
Windows (x64 MSI)	TransEcho_0.1.3_x64_en-US.msi

TransEcho v0.1.2 - Windows 支持

TransEcho v0.1.2

新增 Windows 平台支持！

新功能

• Windows 支持：通过 WASAPI Loopback 捕获系统音频，与 macOS 同等体验
• 音频管线动态适配系统采样率和声道数

改进

• 重命名项目包名 sts → transecho
• 设置界面增加 API 说明和 GitHub 链接
• 目标语言默认跟随系统语言
• 调整最小窗口高度

下载

平台	文件
macOS (Apple Silicon)	TransEcho_0.1.2_aarch64.dmg
Windows (x64)	TransEcho_0.1.2_x64-setup.exe (推荐)
Windows (x64 MSI)	TransEcho_0.1.2_x64_en-US.msi

系统要求

• macOS 14.0+ / Windows 10+
• 豆包同传 API 凭证（免费申请，新用户赠送 100 万 token）

TransEcho v0.1.1

TransEcho v0.1.1

更新内容

• 重命名项目包名 sts → transecho
• 设置界面增加豆包同传 API 说明和快捷跳转链接
• 新用户免费 100 万 token 提示
• 表单标签优化：APP ID / Access Token
• 目标语言默认跟随系统语言
• 设置底部增加 GitHub 仓库链接
• 移除未使用的「中⇄英」语言选项
• 调整最小窗口高度至 650px

下载

• TransEcho_0.1.1_aarch64.dmg - macOS Apple Silicon

TransEcho v0.1.0 - 首个公开版本

TransEcho v0.1.0

macOS 实时同声传译桌面应用，首个公开版本。

功能

• 捕获 macOS 系统音频，实时翻译字幕
• 基于豆包同声传译 2.0 大模型，延迟极低
• 支持语音同传（TTS 播报翻译结果）
• 支持 8 种语言互译：中/英/日/德/法/西/葡/印尼语
• 原生 Rust + Svelte 构建，内存占用极低

系统要求

• macOS 14.0+ (Apple Silicon)
• 需要「屏幕录制」权限
• 需要豆包同传 API 凭证（免费申请，新用户赠送 100 万 token）

安装

1. 下载 TransEcho_0.1.0_aarch64.dmg
2. 打开 DMG，拖拽 TransEcho 到 Applications
3. 首次打开如果提示无法验证开发者，前往 系统设置 → 隐私与安全 → 仍要打开
4. 授予屏幕录制权限
5. 填入 APP ID 和 Access Token，开始使用