AI Voice Chat using Webhook, Memory Manager, OpenAI, Google Gemini & ElevenLabs
工作流概述
这是一个包含15个节点的复杂工作流,主要用于自动化处理各种任务。
工作流源代码
{
"id": "TtoDcjgthgA4NTkU",
"meta": {
"instanceId": "fb261afc5089eae952e09babdadd9983000b3d863639802f6ded8c5be2e40067",
"templateCredsSetupCompleted": true
},
"name": "AI Voice Chat using Webhook, Memory Manager, OpenAI, Google Gemini & ElevenLabs",
"tags": [
{
"id": "mqOrNvCDgQLzPA2x",
"name": "Workflows",
"createdAt": "2024-08-07T14:18:53.614Z",
"updatedAt": "2024-08-07T14:18:53.614Z"
}
],
"nodes": [
{
"id": "86cbf150-df4f-42f7-b7b3-e03c32e6f23c",
"name": "Get Chat",
"type": "@n8n/n8n-nodes-langchain.memoryManager",
"position": [
1700,
-400
],
"parameters": {
"options": {}
},
"typeVersion": 1,
"alwaysOutputData": true
},
{
"id": "a9153a24-e902-4f29-9b83-447317ce3119",
"name": "Insert Chat",
"type": "@n8n/n8n-nodes-langchain.memoryManager",
"position": [
2540,
-400
],
"parameters": {
"mode": "insert",
"messages": {
"messageValues": [
{
"type": "user",
"message": "={{ $('OpenAI - Speech to Text').item.json[\"text\"] }}"
},
{
"type": "ai",
"message": "={{ $json.text }}"
}
]
}
},
"typeVersion": 1,
"alwaysOutputData": true
},
{
"id": "f5c272d4-248b-45a5-87b5-eb659a865d05",
"name": "Sticky Note5",
"type": "n8n-nodes-base.stickyNote",
"position": [
1664,
-491
],
"parameters": {
"color": 6,
"width": 486.4746124819703,
"height": 238.4911357933579,
"content": "## Get Context"
},
"typeVersion": 1
},
{
"id": "32ad17ca-0045-487d-9387-71c2e73629d4",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
2510,
-489
],
"parameters": {
"color": 6,
"width": 321.2536584847704,
"height": 231.05945912581728,
"content": "## Save Context"
},
"typeVersion": 1
},
{
"id": "17ae4f1a-6192-4c52-8157-3cb47b37e0fb",
"name": "Aggregate",
"type": "n8n-nodes-base.aggregate",
"position": [
2020,
-400
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData",
"destinationFieldName": "context"
},
"typeVersion": 1,
"alwaysOutputData": true
},
{
"id": "00b3081e-fbcd-489b-b45a-4e847c346594",
"name": "Window Buffer Memory",
"type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
"position": [
2080,
-100
],
"parameters": {
"sessionKey": "test-0dacb3b5-4bcd-47dd-8456-dcfd8c258204",
"sessionIdType": "customKey"
},
"typeVersion": 1.2
},
{
"id": "55ca2790-e905-414a-a9f6-7d88a9e5807d",
"name": "Google Gemini Chat Model",
"type": "@n8n/n8n-nodes-langchain.lmChatGoogleGemini",
"position": [
2220,
-100
],
"parameters": {
"options": {},
"modelName": "models/gemini-1.5-flash"
},
"credentials": {
"googlePalmApi": {
"id": "2bUF1ZI9hoMIM5XN",
"name": "Google Gemini(PaLM) Api account"
}
},
"typeVersion": 1
},
{
"id": "e8b3433f-b205-404c-9f05-504556d6b6dd",
"name": "Respond to Webhook",
"type": "n8n-nodes-base.respondToWebhook",
"position": [
3560,
-400
],
"parameters": {
"options": {},
"respondWith": "binary"
},
"typeVersion": 1.1
},
{
"id": "de296743-5ac7-454b-bf3a-d020cc024511",
"name": "ElevenLabs - Generate Audio",
"type": "n8n-nodes-base.httpRequest",
"position": [
3240,
-400
],
"parameters": {
"url": "=https://api.elevenlabs.io/v1/text-to-speech/{{voice id}}",
"method": "POST",
"options": {},
"sendBody": true,
"sendHeaders": true,
"authentication": "genericCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "text",
"value": "={{ $('Basic LLM Chain').item.json.text }}"
}
]
},
"genericAuthType": "httpCustomAuth",
"headerParameters": {
"parameters": [
{
"name": "Content-Type",
"value": "application/json"
}
]
}
},
"credentials": {
"httpCustomAuth": {
"id": "lnGfV4BlxSE6Xc4X",
"name": "Eleven Labs"
}
},
"typeVersion": 4.2
},
{
"id": "214e15f2-8a16-4598-b4ac-9fc2ec6545e6",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
3040,
-560
],
"parameters": {
"width": 468.73250812192407,
"height": 843.7602354099661,
"content": "* ### For the Text-to-Speech part, we'll use ElevenLabs.io, which is free and offers a variety of voices to choose from. However, you can also use the OpenAI `\"Generate audio\"` node instead.
* ### Since there is no pre-built node for `\"ElevenLabs\"` in n8n, we'll connect to it through its API using the \"HTTP Request\" node.
## Prerequisites:
* ### `\"ElevenLabs API Key\"` (you can obtain it from their website).
* ### `\"Voice ID\"` (you can also get it from ElevenLabs' \"Voice Library\").
## Setup
* ### In the URL parameter, replace \"{{voice id}}\" at the end of the URL with the Voice ID you obtained from ElevenLabs.io.
* ### To set up your API Key, add custom authentication and include the following `JSON` with your acual ElevenLabs API Key:
```json
{
\"headers\": {
\"xi-api-key\": \"put-your-API-Key-here\"
}
}
```"
},
"typeVersion": 1
},
{
"id": "94ad934c-4a13-47b1-83a5-76fab43b3a47",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
1663,
-598
],
"parameters": {
"color": 6,
"width": 487.4293487597613,
"height": 91.01435855269375,
"content": "### The \"Get Chat,\" \"Insert Chat,\" and \"Window Buffer Memory\" nodes will help the LLM model maintain context throughout the conversation."
},
"typeVersion": 1
},
{
"id": "0a96f48d-0d8b-4240-9eab-a681bfd4c8b5",
"name": "Limit",
"type": "n8n-nodes-base.limit",
"position": [
2900,
-400
],
"parameters": {},
"typeVersion": 1
},
{
"id": "9a5d4ddb-6403-4758-858e-9fbe10c421a9",
"name": "Basic LLM Chain",
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"position": [
2200,
-400
],
"parameters": {
"text": "={{ $('OpenAI - Speech to Text').item.json[\"text\"] }}",
"messages": {
"messageValues": [
{
"type": "AIMessagePromptTemplate",
"message": "=To maintain context and fully understand the user's question, always review the previous conversation between you and him before providing an answer.
This is the previous conversation:
{{ $('Aggregate').item.json[\"context\"].map(m => `
Human: ${m.human || 'undefined'}
AI Assistant: ${m.ai || 'undefined'}
`).join('') }}"
}
]
},
"promptType": "define"
},
"typeVersion": 1.4
},
{
"id": "f2f99895-9678-41b8-ad28-db40e1e23dc0",
"name": "Webhook",
"type": "n8n-nodes-base.webhook",
"position": [
1320,
-400
],
"webhookId": "e9f611eb-a8dd-4520-8d24-9f36deaca528",
"parameters": {
"path": "voice_message",
"options": {},
"httpMethod": "POST",
"responseMode": "responseNode"
},
"typeVersion": 2
},
{
"id": "d9a5fb04-4c02-4da4-b690-7b0ecd0ae052",
"name": "OpenAI - Speech to Text",
"type": "@n8n/n8n-nodes-langchain.openAi",
"position": [
1500,
-400
],
"parameters": {
"options": {},
"resource": "audio",
"operation": "transcribe",
"binaryPropertyName": "voice_message"
},
"credentials": {
"openAiApi": {
"id": "2Cije3KX7OIVwn9B",
"name": "n8n OpenAI"
}
},
"typeVersion": 1.3
}
],
"active": true,
"pinData": {},
"settings": {
"callerPolicy": "workflowsFromSameOwner",
"executionOrder": "v1",
"saveManualExecutions": true
},
"versionId": "fe5792ca-03d7-4cdd-96db-20f4cd479c7e",
"connections": {
"Limit": {
"main": [
[
{
"node": "ElevenLabs - Generate Audio",
"type": "main",
"index": 0
}
]
]
},
"Webhook": {
"main": [
[
{
"node": "OpenAI - Speech to Text",
"type": "main",
"index": 0
}
]
]
},
"Get Chat": {
"main": [
[
{
"node": "Aggregate",
"type": "main",
"index": 0
}
]
]
},
"Aggregate": {
"main": [
[
{
"node": "Basic LLM Chain",
"type": "main",
"index": 0
}
]
]
},
"Insert Chat": {
"main": [
[
{
"node": "Limit",
"type": "main",
"index": 0
}
]
]
},
"Basic LLM Chain": {
"main": [
[
{
"node": "Insert Chat",
"type": "main",
"index": 0
}
]
]
},
"Window Buffer Memory": {
"ai_memory": [
[
{
"node": "Insert Chat",
"type": "ai_memory",
"index": 0
},
{
"node": "Get Chat",
"type": "ai_memory",
"index": 0
}
]
]
},
"OpenAI - Speech to Text": {
"main": [
[
{
"node": "Get Chat",
"type": "main",
"index": 0
}
]
]
},
"Google Gemini Chat Model": {
"ai_languageModel": [
[
{
"node": "Basic LLM Chain",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"ElevenLabs - Generate Audio": {
"main": [
[
{
"node": "Respond to Webhook",
"type": "main",
"index": 0
}
]
]
}
}
}
功能特点
- 自动检测新邮件
- AI智能内容分析
- 自定义分类规则
- 批量处理能力
- 详细的处理日志
技术分析
节点类型及作用
- @N8N/N8N Nodes Langchain.Memorymanager
- Stickynote
- Aggregate
- @N8N/N8N Nodes Langchain.Memorybufferwindow
- @N8N/N8N Nodes Langchain.Lmchatgooglegemini
复杂度评估
配置难度:
维护难度:
扩展性:
实施指南
前置条件
- 有效的Gmail账户
- n8n平台访问权限
- Google API凭证
- AI分类服务订阅
配置步骤
- 在n8n中导入工作流JSON文件
- 配置Gmail节点的认证信息
- 设置AI分类器的API密钥
- 自定义分类规则和标签映射
- 测试工作流执行
- 配置定时触发器(可选)
关键参数
| 参数名称 | 默认值 | 说明 |
|---|---|---|
| maxEmails | 50 | 单次处理的最大邮件数量 |
| confidenceThreshold | 0.8 | 分类置信度阈值 |
| autoLabel | true | 是否自动添加标签 |
最佳实践
优化建议
- 定期更新AI分类模型以提高准确性
- 根据邮件量调整处理批次大小
- 设置合理的分类置信度阈值
- 定期清理过期的分类规则
安全注意事项
- 妥善保管API密钥和认证信息
- 限制工作流的访问权限
- 定期审查处理日志
- 启用双因素认证保护Gmail账户
性能优化
- 使用增量处理减少重复工作
- 缓存频繁访问的数据
- 并行处理多个邮件分类任务
- 监控系统资源使用情况
故障排除
常见问题
邮件未被正确分类
检查AI分类器的置信度阈值设置,适当降低阈值或更新训练数据。
Gmail认证失败
确认Google API凭证有效且具有正确的权限范围,重新进行OAuth授权。
调试技巧
- 启用详细日志记录查看每个步骤的执行情况
- 使用测试邮件验证分类逻辑
- 检查网络连接和API服务状态
- 逐步执行工作流定位问题节点
错误处理
工作流包含以下错误处理机制:
- 网络超时自动重试(最多3次)
- API错误记录和告警
- 处理失败邮件的隔离机制
- 异常情况下的回滚操作