CHIVOX AI

TL;DR — LLMs can't hear audio. Chivox MCP is a hosted MCP server that scores pronunciation at the phoneme level — Mandarin tones included. One tools/call returns overall / accuracy / pron / fluency / details[].phone[] in a stable JSON shape your model can reason over. Not STT. Not a Whisper wrapper.

English | 中文

---

English

🎯 Is this for you?

Most production teams run Whisper + Chivox together: Whisper to transcribe what was said, Chivox to score how well. They don't compete.

What is this?

Chivox Speech Evaluation MCP Server exposes professional speech assessment capabilities as Model Context Protocol (MCP) tools. Connect your AI assistant to our cloud service and let it evaluate pronunciation quality for English and Chinese.

Service endpoint: https://mcp-global.cloud.chivox.com

Features

• 16 evaluation tools — word, sentence, paragraph, phonics, real-time reading, and more
• English + Chinese — 10 English tools, 6 Chinese tools
• Real-time streaming — WebSocket-based live audio evaluation
• Multiple audio inputs — URL, Base64, or file upload
• Works everywhere — Claude Desktop, Cursor, and any MCP-compatible client
• Dual authentication — B2C (API Key) and B2B (JWT)

🚀 Quickstart

Hosted endpoint: https://mcp-global.cloud.chivox.com · every request needs Authorization: Bearer <api_key>. Get a key →

Client	Setup
Cursor	~/.cursor/mcp.json — IDE MCP, zero install
LangChain	LangGraph ReAct agent + MCP adapter
OpenAI Agents SDK	agents.mcp.MCPServerStreamableHttp
Claude Desktop	Local proxy for mic streaming
Raw MCP SDK	Direct mcp Python client

Cursor (zero install)

// ~/.cursor/mcp.json
{
  "mcpServers": {
    "chivox-speech-eval": {
      "type": "streamable-http",
      "url": "https://mcp-global.cloud.chivox.com",
      "headers": { "Authorization": "Bearer <your_api_key>" }
    }
  }
}

LangChain

from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent

client = MultiServerMCPClient({
    "chivox": {
        "transport": "streamable_http",
        "url": "https://mcp-global.cloud.chivox.com",
        "headers": {"Authorization": "Bearer <your_api_key>"},
    }
})
tools = await client.get_tools()  # discovers all 16 tools

agent = create_react_agent("openai:gpt-4o-mini", tools)
result = await agent.ainvoke({"messages": [(
    "user",
    "Score https://example.com/audio/sentence.mp3, ref: I think therefore I am",
)]})

OpenAI Agents SDK

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

chivox = MCPServerStreamableHttp(
    params={
        "url": "https://mcp-global.cloud.chivox.com",
        "headers": {"Authorization": "Bearer <your_api_key>"},
    },
    name="chivox-speech-eval",
)

async with chivox:
    agent = Agent(
        name="coach",
        instructions="Professional speaking coach",
        mcp_servers=[chivox],
    )
    r = await Runner.run(
        agent,
        "Score https://example.com/audio/sentence.mp3, ref: I think therefore I am",
    )
    print(r.final_output)

Claude Desktop (mic streaming via local proxy)

npm install -g chivox-local-mcp

// ~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "chivox": {
      "command": "chivox-local-mcp",
      "env": {
        "MCP_REMOTE_URL": "https://mcp-global.cloud.chivox.com",
        "MCP_API_KEY": "<your_api_key>"
      }
    }
  }
}

Raw MCP SDK

import asyncio
from mcp.client.streamable_http import streamablehttp_client
from mcp import ClientSession

async def main():
    async with streamablehttp_client(
        "https://mcp-global.cloud.chivox.com",
        headers={"Authorization": "Bearer <your_api_key>"},
    ) as (r, w, _):
        async with ClientSession(r, w) as s:
            await s.initialize()
            out = await s.call_tool("en_sentence_eval", {
                "ref_text": "I think therefore I am",
                "audio_url": "https://example.com/audio/sentence.mp3",
            })
            print(out)

asyncio.run(main())

More clients (Claude Code, Windsurf, Zed, Mastra, function-calling mode) → docs → Clients

🧠 What the LLM actually sees

Every tool returns the same top-level shape — switch locale or granularity with zero schema work. Example for "hello":

{
  "overall": 85,
  "accuracy": 82,
  "pron": 88,
  "integrity": 95,
  "fluency": { "overall": 78, "speed": 65, "pause": 2 },
  "details": [
    {
      "char": "hello",
      "score": 85,
      "phone": [
        { "phoneme": "h",  "score": 90, "dp_type": "normal" },
        { "phoneme": "ɛ",  "score": 82, "dp_type": "normal" },
        { "phoneme": "l",  "score": 88, "dp_type": "normal" },
        { "phoneme": "oʊ", "score": 80, "dp_type": "normal" }
      ]
    }
  ]
}

For English mispronunciations, phoneme_error: { expected, actual } is included. Mandarin adds tone_ref / tone_detected with sandhi-aware dp_type verdicts. Full field list →

🛠️ Tools catalog

Inline audio: pass audio_url or audio_base64 in the tool call — no upload round-trip. Formats: mp3 · wav · ogg · m4a · aac · pcm.

English Evaluation

Tool	Description	core_type
en_word_eval	Word pronunciation scoring	en.word.score
en_word_correction	Word pronunciation correction	en.word.pron
en_phonics_eval	Phonics evaluation	en.nsp.score
en_sentence_eval	Sentence reading assessment	en.sent.score
en_sentence_correction	Sentence pronunciation correction	en.sent.pron
en_vocab_eval	Multi-word evaluation	en.vocabs.pron
en_paragraph_eval	Paragraph reading assessment	en.pred.score
en_realtime_eval	Real-time reading evaluation	en.rltm.score
en_choice_eval	Oral choice evaluation	en.choc.score
en_semi_open_eval	Semi-open question evaluation	en.scne.exam

Chinese Evaluation

Tool	Description	core_type
cn_word_pinyin_eval	Pinyin pronunciation scoring	cn.word.score
cn_word_raw_eval	Character pronunciation scoring	cn.word.raw
cn_sentence_eval	Sentence reading assessment	cn.sent.raw
cn_paragraph_eval	Paragraph reading assessment	cn.pred.raw
cn_rec_eval	Limited-branch recognition	cn.rec.raw
cn_aitalk_eval	AI Talk — oral expression evaluation	cn.recscore.raw

Streaming Evaluation

Tool	Description
create_stream_session	Create a streaming session, returns session_id and WebSocket URL

🔌 Dual transport

Two ways to feed audio — same result shape, different UX. Function-calling fallback: fc-global.cloud.chivox.com.

⚖️ How it compares

Rule of thumb — use Whisper to know what was said; use Chivox to know how well. They stack.

💬 …and here's what your LLM does with it

Pipe that JSON straight into any chat model with a one-line system prompt — "You are a warm pronunciation coach. Diagnose, then drill." — and you get a real lesson back. No fine-tuning. No audio understanding. Just chat.completion.

Why this works — the LLM never "heard" the audio. The JSON names the problem in fields it already understands (dp_type: "mispron", phoneme_error.actual, tone_ref vs tone_detected), so a vanilla chat.completion can diagnose like a human teacher.

🔁 The three-stage loop

🎤 Input: 1-minute learner recording → Output: warm feedback + targeted drill, end-to-end in < 1.6 seconds.

🏮 The moat: a tireless Mandarin tutor

30M+ learners worldwide study Mandarin — including heritage speakers and adult beginners — yet few platforms score tone errors (mā / má / mǎ / mà) at the phoneme level in English. Chivox's Chinese engine is trained on the same data that powers China's Putonghua Proficiency Test (普通话水平测试, PSC).

🇬🇧 And yes — exam-grade English too

Exam-grade rubrics on the same MCP endpoints: IELTS · TOEFL · Cambridge YLE · K-12 reading assessments for English, plus PSC-aligned Mandarin scoring. Same JSON shape, 20+ scoring dimensions — just change ref_text and accent.

💎 Why developers ship with Chivox MCP

Plus: streaming + inline modes · TLS 1.3 end-to-end · audio discarded after scoring (JSON retained 30 days) · on-prem available for enterprise · limits & privacy →

💳 Pricing

Honest defaults. Start with 600 free calls (30 days) and all 16 tools unlocked — no feature gates, no card. When you need more, pay per successful call at tiered rates — the more you ship, the cheaper each call gets.

Free tier ≠ crippled tier. Every new account gets 600 free calls valid for 30 days with the full 16-tool catalog — same engine, same JSON, same SLA as paid keys. After the trial window or when calls are used up, top up from $10 and let the volume tiers do the rest. Failed calls are never billed.

❓ FAQ

Is this just another wrapper around Whisper? No. Whisper transcribes; Chivox scores. The engine is trained on exam-graded samples and returns phoneme-level details[].phone[] — not a transcript. Most teams run both.

Does it work offline / on-device? The hosted MCP server needs outbound access to the scoring engine. For air-gapped deployments, contact us — we ship an on-prem container for enterprise customers.

What about dialects and accents? Mandarin targets standard Pǔtōnghuà with sandhi-aware tone verdicts. English supports en-US, en-GB, and en-AU rubrics via locale parameters on the relevant tools.

Which LLMs work out of the box? Any model with OpenAI-style function calling: GPT-4o / 5.x, Claude Sonnet / Opus, Gemini, DeepSeek, GLM, Kimi, Doubao, Qwen. Tool schemas are forwarded verbatim.

Can I use this in a browser? For quick demos, yes — but production traffic should flow through your backend so the API key stays server-side. Privacy notes →

Streaming Workflow

1. Create session    →  tools/call: create_stream_session
                         ↓ returns ws_url
2. Connect WebSocket →  wss://{ws_url}/ws/audio/{session_id}
                         ↓
3. Send audio frames →  Binary frames (8KB chunks recommended)
                         ↓
4. Stop & get result →  Send {"cmd": "stop"}, receive final scores

Examples

Example	Description
Customer Proxy Server	Python Flask server that securely proxies MCP calls (keeps API Key server-side)
Claude Desktop Config	Configuration guide for Claude Desktop
Cursor Config	Configuration guide for Cursor
Quick Test Scripts	Minimal test scripts in Python, Node.js, and curl

Documentation

• API Reference — Full protocol documentation, authentication, tool parameters, and streaming details

Support

• Website: chivox.com
• Issues: GitHub Issues

---

中文

🎯 适合场景

多数生产团队 Whisper + 驰声一起用：Whisper 识别说了什么，驰声评估说得怎么样。两者互补。

简介

驰声语音评测 MCP 服务基于 Model Context Protocol (MCP) 标准，将专业语音评测能力封装为 MCP 工具，供 AI 客户端（Claude Desktop、Cursor 等）直接调用。

服务地址： https://mcp-global.cloud.chivox.com

功能特性

• 16 种评测工具 — 单词、句子、段落、自然拼读、实时朗读等
• 中英文双语 — 10 种英文评测 + 6 种中文评测
• 实时流式评测 — 通过 WebSocket 实时推送音频，获取评测结果
• 多种音频输入 — 支持 URL、Base64 编码、文件上传
• 广泛兼容 — 支持 Claude Desktop、Cursor 及任何 MCP 兼容客户端
• 双认证模式 — B2C（API Key）和 B2B（JWT 签名）

🚀 快速开始

服务地址：https://mcp-global.cloud.chivox.com · 每个请求需携带 Authorization: Bearer <api_key>。获取 Key →

客户端	接入方式
Cursor	~/.cursor/mcp.json — IDE MCP，零安装
LangChain	LangGraph ReAct agent + MCP 适配器
OpenAI Agents SDK	agents.mcp.MCPServerStreamableHttp
Claude Desktop	本地代理，支持麦克风流式传输
Raw MCP SDK	直接使用 mcp Python 客户端

Cursor (零安装)

// ~/.cursor/mcp.json
{
  "mcpServers": {
    "chivox-speech-eval": {
      "type": "streamable-http",
      "url": "https://mcp-global.cloud.chivox.com",
      "headers": { "Authorization": "Bearer <your_api_key>" }
    }
  }
}

LangChain

from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent

client = MultiServerMCPClient({
    "chivox": {
        "transport": "streamable_http",
        "url": "https://mcp-global.cloud.chivox.com",
        "headers": {"Authorization": "Bearer <your_api_key>"},
    }
})
tools = await client.get_tools()  # 自动发现全部 16 个工具

agent = create_react_agent("openai:gpt-4o-mini", tools)
result = await agent.ainvoke({"messages": [(
    "user",
    "评测 https://example.com/audio/sentence.mp3，参考文本：I think therefore I am",
)]})

OpenAI Agents SDK

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

chivox = MCPServerStreamableHttp(
    params={
        "url": "https://mcp-global.cloud.chivox.com",
        "headers": {"Authorization": "Bearer <your_api_key>"},
    },
    name="chivox-speech-eval",
)

async with chivox:
    agent = Agent(
        name="coach",
        instructions="专业口语教练",
        mcp_servers=[chivox],
    )
    r = await Runner.run(
        agent,
        "评测 https://example.com/audio/sentence.mp3，参考文本：I think therefore I am",
    )
    print(r.final_output)

Claude Desktop （本地代理 + 麦克风流式传输）

npm install -g chivox-local-mcp

// ~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "chivox": {
      "command": "chivox-local-mcp",
      "env": {
        "MCP_REMOTE_URL": "https://mcp-global.cloud.chivox.com",
        "MCP_API_KEY": "<your_api_key>"
      }
    }
  }
}

Raw MCP SDK

import asyncio
from mcp.client.streamable_http import streamablehttp_client
from mcp import ClientSession

async def main():
    async with streamablehttp_client(
        "https://mcp-global.cloud.chivox.com",
        headers={"Authorization": "Bearer <your_api_key>"},
    ) as (r, w, _):
        async with ClientSession(r, w) as s:
            await s.initialize()
            out = await s.call_tool("en_sentence_eval", {
                "ref_text": "I think therefore I am",
                "audio_url": "https://example.com/audio/sentence.mp3",
            })
            print(out)

asyncio.run(main())

更多客户端（Claude Code、Windsurf、Zed、Mastra、function-calling 模式）→ 文档 → 客户端

🧠 LLM 看到的数据

每个工具返回 统一的顶层结构 — 切换语种或粒度无需修改 schema。以 "hello" 为例：

{
  "overall": 85,
  "accuracy": 82,
  "pron": 88,
  "integrity": 95,
  "fluency": { "overall": 78, "speed": 65, "pause": 2 },
  "details": [
    {
      "char": "hello",
      "score": 85,
      "phone": [
        { "phoneme": "h",  "score": 90, "dp_type": "normal" },
        { "phoneme": "ɛ",  "score": 82, "dp_type": "normal" },
        { "phoneme": "l",  "score": 88, "dp_type": "normal" },
        { "phoneme": "oʊ", "score": 80, "dp_type": "normal" }
      ]
    }
  ]
}

英文错误发音会包含 phoneme_error: { expected, actual }。中文额外提供 tone_ref / tone_detected 及变调感知的 dp_type 判定。完整字段列表 →

🛠️ 工具一览

内联音频： 在工具调用中直接传入 audio_url 或 audio_base64 — 无需额外上传。支持格式： mp3 · wav · ogg · m4a · aac · pcm。

英文评测

工具名	说明	core_type
en_word_eval	单词评测 — 总分及每个音标得分	en.word.score
en_word_correction	单词纠音 — 发音纠正建议	en.word.pron
en_phonics_eval	自然拼读评测	en.nsp.score
en_sentence_eval	句子评测 — 流利度、准确度、完整度	en.sent.score
en_sentence_correction	句子纠音	en.sent.pron
en_vocab_eval	词语评测	en.vocabs.pron
en_paragraph_eval	段落评测 — 每句、每词得分	en.pred.score
en_realtime_eval	实时朗读评测	en.rltm.score
en_choice_eval	口语选择题评测	en.choc.score
en_semi_open_eval	半开放题评测	en.scne.exam

中文评测

工具名	说明	core_type
cn_word_pinyin_eval	拼音评测 — 总分、声母、韵母、声调	cn.word.score
cn_word_raw_eval	汉字评测	cn.word.raw
cn_sentence_eval	词句评测 — 总分、声调、准确度、流利度	cn.sent.raw
cn_paragraph_eval	段落评测	cn.pred.raw
cn_rec_eval	有限分支识别	cn.rec.raw
cn_aitalk_eval	AI Talk — 识别并评测口语表达	cn.recscore.raw

流式评测

工具名	说明
create_stream_session	创建流式评测会话，返回 session_id 和 WebSocket 地址

🔌 双传输模式

两种传输音频的方式 — 返回结构一致，体验不同。Function-calling 备选地址：fc-global.cloud.chivox.com。

⚖️ 方案对比

经验法则 — 用 Whisper 知道说了什么；用 驰声 知道说得怎么样。两者可叠加使用。

💬 LLM 如何利用评测结果

将 JSON 直接传给任何对话模型，配上一句系统提示 — "你是一位温暖的发音教练，先诊断再练习。" — 就能得到真正的教学反馈。无需微调，无需音频理解，只需 chat.completion。

为什么可行 — LLM 从未"听到"音频。JSON 用模型已理解的字段命名了问题（dp_type: "mispron"、phoneme_error.actual、tone_ref vs tone_detected），因此原生 chat.completion 就能像真人教师一样诊断。

🔁 三阶段闭环

🎤 输入： 1 分钟学习者录音 → 输出： 温暖反馈 + 针对性练习，端到端 < 1.6 秒。

🏮 核心优势：不知疲倦的中文普通话导师

全球超过 3000 万 学习者在学习普通话 — 包括华裔传承语使用者和成人初学者 — 但很少有平台能在音素级别评测声调错误（mā / má / mǎ / mà）。驰声的中文引擎使用与中国普通话水平测试（PSC）相同的数据训练。

🇬🇧 同样出色的考试级英文评测

同一 MCP 端点提供考试级评分标准：英文支持 IELTS · TOEFL · Cambridge YLE · K-12 朗读评测，中文对齐 PSC 评分。统一 JSON 结构，20+ 评分维度 — 只需更换 ref_text 和 accent。

💎 开发者选择驰声 MCP 的理由

另外：流式 + 内联 双模式 · TLS 1.3 全链路加密 · 评测后音频即删（JSON 保留 30 天）· 企业客户可私有部署 · 限制与隐私 →

💳 定价

实在的默认策略。注册即获 600 次免费调用（30 天有效），全部 16 个工具解锁 — 无功能限制，无需绑卡。需要更多时，按成功调用计费，阶梯费率 — 用得越多，单价越低。

免费版 ≠ 阉割版。 每个新账户获得 600 次免费调用（30 天有效），完整 16 工具目录 — 引擎、JSON、SLA 与付费版完全一致。试用到期或用完后，最低 $10 充值，享受 阶梯优惠。失败调用不计费。

❓ 常见问题

这是不是又一个 Whisper 封装？ 不是。Whisper 做语音转文字，驰声做发音评分。引擎基于考试评分样本训练，返回音素级 details[].phone[] — 不是转录文本。大多数团队两者并用。

支持离线/端侧部署吗？ 托管 MCP 服务需要连接评分引擎。如需气隙部署，请联系我们 — 我们为企业客户提供私有化容器。

支持方言和口音吗？ 中文目标为标准普通话，支持变调感知的声调判定。英文通过工具参数支持 en-US、en-GB、en-AU 评分标准。

哪些 LLM 可以直接使用？ 任何支持 OpenAI 风格 function calling 的模型：GPT-4o / 5.x、Claude Sonnet / Opus、Gemini、DeepSeek、GLM、Kimi、豆包、Qwen。工具 schema 原样透传。

可以在浏览器中使用吗？ 快速演示可以，但生产环境请通过后端转发，确保 API Key 保存在服务端。隐私说明 →

流式评测流程

1. 创建会话      →  tools/call: create_stream_session
                      ↓ 返回 ws_url
2. 连接 WebSocket →  wss://{ws_url}/ws/audio/{session_id}
                      ↓
3. 推送音频帧     →  二进制帧（建议每帧 8KB）
                      ↓
4. 停止并获取结果 →  发送 {"cmd": "stop"}，接收最终评分

示例

示例	说明
客户代理服务器	Python Flask 代理服务，API Key 安全中转
Claude Desktop 配置	Claude Desktop 接入指南
Cursor 配置	Cursor 接入指南
快速测试脚本	Python、Node.js、curl 最小化调用示例

文档

• API 接入文档 — 完整协议文档、认证方式、工具参数、流式评测说明

支持

• 官网：chivox.com
• 问题反馈：GitHub Issues

---

🤝 Star us · say hi

---

License

The example code and documentation in this repository are licensed under the MIT License.

The Chivox Speech Evaluation MCP Service itself is a commercial product. Visit the Chivox API Portal for service access and pricing.