Bilingual Mix

io.github.zxm9988由 GitHub 账号 zxm9988 发布v1.0.0更新于 Nov 8, 2026

Remote MCP server for sentence-by-sentence alternating bilingual listening audio.

已验证Streamable HTTP可网页运行媒体与设计AI 与机器学习

概览

AI 生成的概览

远程 MCP 服务器，把中英或日英对照句子合成为逐句交替的双语听力音频。

可以做什么
  • 生成双语音频：把成对句子用神经语音合成一条 A→B 交替的练习音轨，可设置间隔，返回 WAV 下载链接和逐句时间轴
  • 对齐已有录音：用 Whisper 转写两个公开音频链接，并按语义配对成带时间戳和相似度的句子对
  • 翻译文本行：在中英之间翻译一组文本行，保持顺序和行数（会员功能）
  • 查询额度与套餐：登录状态、已用与剩余分钟、单次上限和价格档位
适用场景
适合让助手为语言学习者制作逐句交替的双语听力材料，无论是已有对照文本还是两段现成录音；也适合在消耗额度前先确认剩余分钟数。
运行要求
使用 bilingualmix.com 上的远程 Streamable HTTP 端点，无需本地安装，未声明 API 密钥或请求头。匿名调用有按 IP 计算的免费额度，用完后服务器返回 401，客户端需通过 OAuth 2.1（PKCE）使用已有的 bilingualmix.com 账号登录。align_audio 需要两个可公开访问的音频链接。
安装前请注意
调用会消耗按次数和分钟计量的额度，并存在付费会员档位，建议先用 dry_run 预览成本再生成。匿名额度按 IP 计算并共享。传给 align_audio 的音频链接会由服务端抓取，README 称原始音频和转写文本不会被保存。登录会通过 OAuth 授权访问账号。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Bilingual Mix，将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Web executable，通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体，或 SourceWeft 对话中。智能体会读取该服务器的安装指南，先向你展示发布者、端点和登录方式，经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/io.github.zxm9988%2Fbilingual-mix/install.md，并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前，先告诉我它的发布者、端点和登录方式，等我确认。不要把 API 密钥写进命令或我会提交的文件；问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。

{
  "mcpServers": {
    "bilingual-mix": {
      "type": "http",
      "url": "https://bilingualmix.com/mcp"
    }
  }
}

README

Bilingual Mix

Turn matching Mandarin or Japanese and English audio into one sentence-by-sentence listening track.

Try Bilingual Mix →

Bilingual Mix is a browser-based listening tool for language learners. Bring an audiobook, podcast or lesson in Mandarin or Japanese together with its English version. The tool aligns corresponding sentences, lets you choose language order and pauses, and exports a standard MP3 for listening on any device.

Highlights

  • Sentence-by-sentence bilingual listening
  • Mandarin and Japanese learning workflows
  • MP3 export
  • No sign-up needed for the free trial
  • Raw audio and transcripts are not stored by the service

Remote MCP server

Bilingual Mix also runs as a remote Model Context Protocol server, so AI agents can build alternating-language practice audio directly.

  • Endpoint: https://bilingualmix.com/mcp (Streamable HTTP, stateless, protocol version 2025-06-18)
  • Landing page: English · 中文 · 日本語
  • Machine-readable description: server.json

For MCP clients that accept remote HTTP servers:

json
{  "mcpServers": {    "bilingual-mix": {      "type": "streamable-http",      "url": "https://bilingualmix.com/mcp"    }  }}

Tools:

ToolWhat it does
create_bilingual_audioSentence pairs to one A→B alternating track, with a sentence timeline. dry_run: true quotes the cost without spending quota.
align_audioTwo audio URLs to transcribed, cross-language aligned sentence pairs with timestamps.
translate_linesFills in the other side when only one language's text is available (members).
get_quotaSign-in state, minutes used and remaining, per-run limits.
list_plansPlans and quotas.

Anonymous calls get a free per-IP allowance (2 runs and 20 minutes a day, up to 10 minutes and 8 sentence pairs per run). When it runs out, the server replies 401 with a WWW-Authenticate resource metadata challenge; clients then sign in through OAuth 2.1 with PKCE (S256), including dynamic client registration, to use their account quota. Discovery documents are at /.well-known/oauth-protected-resource and /.well-known/oauth-authorization-server. A signed-in free account gets 4 runs and 40 minutes a day; a membership gets 500 minutes a month, up to 45 minutes and 30 sentence pairs per run.

Learn more

[Bilingual Mix - Featured on BoringCashCow]

来源：README.md，提交 787bcea

工具

5

create_bilingual_audio

Create bilingual interleaved audio

Turn parallel sentences (language A + language B) into one alternating practice track. Each sentence is spoken by the same neural voices used on bilingualmix.com, then stitched A1→B1→A2→B2 with configurable gaps. Returns a 6-hour download URL (WAV) plus a per-sentence timeline in seconds. Use dry_run=true first to preview the pair list and the cost without spending quota. Free tier: 2 runs / 20 min per day (anonymous, per IP), up to 8 pairs. Signing in doubles it; members get 500 min/month.

输入参数textpairsgap_mslang_alang_btext_bdry_runpair_gap_ms

align_audio

Align two audio files into sentence pairs

Given two publicly reachable audio URLs (same content in two languages), transcribes both with Whisper and aligns them semantically into sentence pairs with timestamps and a similarity score. Use this when the user already has recordings rather than text. Free tier limits apply (2 runs / 20 min per day anonymous, ≤10 min combined per run).

输入参数lang_alang_baudio_a_url必填audio_b_url必填

translate_lines

Translate parallel lines

Translate a list of lines between Chinese and English, keeping order and line count. Member feature — anonymous and free accounts get an auth prompt. For free usage, pass both sides yourself to create_bilingual_audio instead.

输入参数fromlines必填

get_quota

Check remaining quota

Report sign-in state, plan, minutes used and remaining, and per-run limits. Free, unlimited.

无需输入参数。

list_plans

List pricing plans

Pricing tiers, minute allowances and per-run limits. Free, unlimited.

无需输入参数。

版本历史

1
  1. v1.0.0最新Nov 8, 2026

评分

已满足适用检查项的 90%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通必需25 分
  • 已声明认证方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分

    请在仓库或注册中心条目中声明许可证。