没有合适的资源?快使用搜索试试~ 我知道了~
首页阿里智能语音V2.X SDM(MRCP-SERVER)技术文档(1).pdf
阿里智能语音V2.X SDM(MRCP-SERVER)技术文档(1).pdf
需积分: 48 2.0k 浏览量
更新于2023-03-16
评论
收藏 1.74MB PDF 举报
ASR 语音识别技术,也称为自动语音识别(Automatic Speech Recognition),简称 ASR,其目 标是将人类语音中的词汇内容转换为可读的文字。 TTS 语音合成技术,也称为自动语音合成(Text To Speech),简称 TTS,其目标是将文字转 成对应的语音声音。 NLU 自然语言理解技术(Natural Language Understanding),简称 NLU,有的叫做自然语言 处理(Natural Language Processing, NLP), 这里认为这两者是同一个概念,即:研究如 何让计算机读懂人类语言。 IVR 交互式语音应答技术(Interactive Voice Response),简称 IVR,本文将呼叫中心(Call Center)统一概称为 IVR。一般来说,由 IVR 通过 SDM 服务(实现了 MRCP 协议)调用 ASR、TTS、NLU 能力。 MRCP-SERVER 语音对话管理服务(Speech Dialogue Managerment),简称 SDM,也即是本文档所描述的 服务,是 MRCP 协议的服务端实现,对外用以和各类呼叫平台(比如华为呼叫中心、 avaya、freeswitch)进行对接,对内集成了
资源详情
资源评论
资源推荐

2019-3-5
SDM(MRCP-SERVER)技术文档

2
目录
1.
名词解释 ................................................................................................ 6
2.
概述 ....................................................................................................... 7
2.1 背景 ...................................................................................................... 7
2.2 MRCP 协议 ............................................................................................ 7
2.3 MRCP-SERVER ..................................................................................... 8
3.
SDM 核心功能列表 ................................................................................. 9
4.
语音合成(TTS)接口 ............................................................................... 10
4.1 MRCP 特性 .......................................................................................... 10
4.1.1 MRCP 方法 ......................................................................................................... 10
4.1.2 MRCP 事件 ......................................................................................................... 10
4.1.3 MRCP 消息头 ...................................................................................................... 10
4.2 MRCP 返回码 ...................................................................................... 11
4.3 MRCP 状态机 ...................................................................................... 12
4.4 MRCP 请求 .......................................................................................... 12
5.
语音识别(ASR)接口 ............................................................................... 14
5.1 MRCP 特性 ............................................................................................... 14
5.1.1 MRCP 方法 ......................................................................................................... 14
5.1.2 MRCP 事件 ......................................................................................................... 14
5.1.3 MRCP 消息头 ...................................................................................................... 14

3
5.2 MRCP 返回码 ............................................................................................ 15
5.3 MRCP 状态机 ............................................................................................ 16
5.4 MRCP 请求 ............................................................................................... 16
5.5 Vendor-Specific-Parameters 参数 ................................................................... 18
5.5.1 ASR 识别参数 ..................................................................................................... 18
5.5.2 业务自定义参数 .................................................................................................. 19
6.
部署 ..................................................................................................... 20
6.1 全私有云部署........................................................................................ 20
6.2 私有云+公共云混合部署 ........................................................................ 21
6.2.1 开通阿里云智能语音 ASR、TTS 服务 ................................................................. 21
6.2.2 配置 SDM(MRCP-SERVER) ................................................................................. 22
7.
集成对接 .............................................................................................. 24
7.1 快速入门 .............................................................................................. 24
7.1.1 什么是 MRCP 协议,用来做什么 ........................................................................ 24
7.1.2 和 IVR(呼叫中心)对接时需要准备什么 ............................................................... 24
7.1.3 SDM 使用的端口及协议都有哪些 ....................................................................... 25
7.1.4 支持哪些语音格式 .............................................................................................. 25
7.1.5 返回的 ASR 结果格式是什么样的 ....................................................................... 26
7.1.6 为什么有的 IVR 需要语法文件............................................................................ 26
7.1.7 如何调整无话超时参数 ....................................................................................... 27
7.1.8 no-match 和 no-input-timeout 是什么意思 ............................................................. 27

4
7.1.9 如何实现语音打断 .............................................................................................. 28
7.2 和 ASR 服务的联动 ................................................................................ 29
7.2.1 ASR 相关参数配置 .............................................................................................. 29
7.2.2 VAD 断句间隔时长(静默时长) ............................................................................ 30
7.2.3 ASR (泛)热词 ....................................................................................................... 31
7.2.4 ASR 类热词 ......................................................................................................... 31
7.2.5 ASR 定制语言模型 .............................................................................................. 31
7.2.6 是否需要对 ASR 文本加标点 ............................................................................... 32
7.2.7 是否需要对 ASR 文本进行规整/顺滑 .................................................................. 32
7.3 和 TTS 服务的联动 ................................................................................ 32
7.3.1 TTS 发音人设置 .................................................................................................. 33
7.3.2 TTS 音量调整 ...................................................................................................... 33
7.3.3 TTS 语速调整 ...................................................................................................... 34
7.3.4 TTS 语调调整 ...................................................................................................... 34
8.
常见示例流程 ....................................................................................... 35
8.1 TTS .......................................................................................................... 35
8.1.1 IVR 发送纯文本 .................................................................................................. 35
8.1.2 IVR 发送带 SSML 标签的文本 ............................................................................. 35
8.2 ASR .......................................................................................................... 36
8.2.1 先发送 Define-Grammar 再发送 Recognize ............................................................ 36
8.2.2 直接发送 Recognize ............................................................................................. 37

5
8.2.3 先发送 Recognize 再发送 Start-Input-Timers .......................................................... 37
8.2.4 连续语音识别请求 .............................................................................................. 38
9.
问题排查 .............................................................................................. 42
9.1 日志说明 .............................................................................................. 42
9.1.1 alimrcp-server.log .................................................................................................. 42
9.1.2 sdm-asr-request.log ................................................................................................ 42
9.1.3 sdm-tts-request.log ................................................................................................ 43
9.2 错误码快查 ........................................................................................... 44
9.2.1 ASR 请求相关错误 .............................................................................................. 44
9.2.2 TTS 请求相关错误 .............................................................................................. 45
9.3 语音识别率低的排查 ............................................................................. 45
9.4 识别延迟高,识别慢的排查 ................................................................... 46
9.5 语音打断慢,延迟大的排查 ................................................................... 47
9.6 其他常见错误........................................................................................ 49
9.6.1 DNS 解析错误 .................................................................................................... 49
剩余48页未读,继续阅读

















安全验证
文档复制为VIP权益,开通VIP直接复制

评论0