Merge branch 'feat/ivr-lang' into develop

microsoft · Feb 9, 2024 · 131e60c · 131e60c
2 parents 19358ae + 3ad1a7c
commit 131e60c
Show file tree

Hide file tree

Showing 22 changed files with 670 additions and 329 deletions.
diff --git a/README.md b/README.md
@@ -47,6 +47,7 @@ Extract of the data stored during the call:
 
 - [x] Access the claim on a public website
 - [x] Access to customer conversation history
+- [x] Allow user to change the language of the conversation
 - [x] Bot can be called from a phone number
 - [x] Bot use multiple voice tones (e.g. happy, sad, neutral) to keep the conversation engaging
 - [x] Company products (= lexicon) can be understood by the bot (e.g. a name of a specific insurance product)
@@ -106,9 +107,11 @@ graph LR
     db[("Conversations and claims\n(Cosmos DB or SQLite)")]
     event_grid[("Broker\n(Event Grid)")]
     gpt["GPT-4 Turbo\n(OpenAI)"]
+    translation["Translation\n(Cognitive Services)"]
   end
 
   api -- Answer with text --> communication_service
+  api -- Ask for translation --> translation
   api -- Few-shot training --> ai_search
   api -- Generate completion --> gpt
   api -- Save conversation --> db
@@ -211,6 +214,7 @@ workflow:
   agent_phone_number: "+33612345678"
   bot_company: Contoso
   bot_name: Robert
+  lang: {}
 
 communication_service:
   phone_number: "+33612345678"
@@ -264,11 +268,9 @@ communication_service:
   access_key: xxx
   endpoint: https://xxx.france.communication.azure.com
   phone_number: "+33612345678"
-  voice_name: fr-FR-DeniseNeural
 
 cognitive_service:
   # Must be of type "AI services multi-service account"
-  # See: https://learn.microsoft.com/en-us/azure/ai-services/multi-service-resource?tabs=macos&pivots=azportal#create-a-new-multi-service-resource
   endpoint: https://xxx.cognitiveservices.azure.com
 
 openai:
@@ -373,7 +375,7 @@ prompts:
       Assistant will provide internal IT support to employees. Assistant requires data from the employee to provide IT support. The assistant's role is not over until the issue is resolved or the request is fulfilled.
 
       # Rules
-      - Answers in {conversation_lang}, even if the employee speaks in a different language
+      - Answers in {default_lang}, even if the customer speaks another language
       - Cannot talk about any topic other than IT support
       - Is polite, helpful, and professional
       - Rephrase the employee's questions as statements and answer them
@@ -401,6 +403,28 @@ prompts:
       {reminders}
 ```
 
+### Customize the languages
+
+The bot can be used in multiple languages. It can understand the language the user chose.
+
+See the [list of supported languages](https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=tts#supported-languages) for the Text-to-Speech service.
+
+```yaml
+# config.yaml
+[...]
+
+workflow:
+  lang:
+    default_short_code: "fr-FR"
+    availables:
+      - pronunciations_en: ["French", "FR", "France"]
+        short_code: "fr-FR"
+        voice_name: "fr-FR-DeniseNeural"
+      - pronunciations_en: ["Chinese", "ZH", "China"]
+        short_code: "zh-CN"
+        voice_name: "zh-CN-XiaoxiaoNeural"
+```
+
 ### Customize the moderation levels
 
 Levels are defined for each category of Content Safety. The higher the score, the more strict the moderation is, from 0 to 7.

diff --git a/bicep/app.bicep b/bicep/app.bicep
@@ -4,7 +4,6 @@ param agentPhoneNumber string
 param botCompany string
 param botName string
 param botPhoneNumber string
-param botVoiceName string
 param gptModel string
 param gptVersion string
 param imageVersion string
@@ -38,12 +37,12 @@ var config = {
     agent_phone_number: agentPhoneNumber
     bot_company: botCompany
     bot_name: botName
+    lang: loadYamlContent('../config.yaml').workflow.lang
   }
   communication_service: {
     access_key: communication.listKeys().primaryKey
     endpoint: communication.properties.hostName
     phone_number: botPhoneNumber
-    voice_name: botVoiceName
   }
   cognitive_service: {
     endpoint: cognitiveCommunication.properties.endpoint
@@ -68,6 +67,10 @@ var config = {
     llm: loadYamlContent('../config.yaml').prompts.llm
     tts: loadYamlContent('../config.yaml').prompts.tts
   }
+  ai_translation: {
+    access_key: translate.listKeys().key1
+    endpoint: 'https://${translate.name}.cognitiveservices.azure.com/'
+  }
 }
 
 output appUrl string = appUrl
@@ -81,7 +84,7 @@ resource logAnalyticsWorkspace 'Microsoft.OperationalInsights/workspaces@2022-10
   properties: {
     retentionInDays: 30
     sku: {
-      name: 'PerGB2018'
+      name: 'PerGB2018'  // Pay-as-you-go
     }
   }
 }
@@ -221,7 +224,7 @@ resource cognitiveCommunication 'Microsoft.CognitiveServices/accounts@2023-10-01
   location: location
   tags: tags
   sku: {
-    name: 'S0'
+    name: 'S0'  // Only one available
   }
   kind: 'CognitiveServices'
   properties: {
@@ -234,7 +237,7 @@ resource cognitiveDocument 'Microsoft.CognitiveServices/accounts@2023-10-01-prev
   location: location
   tags: tags
   sku: {
-    name: 'S0'
+    name: 'S0'  // Pay-as-you-go
   }
   kind: 'FormRecognizer'
   properties: {
@@ -247,7 +250,7 @@ resource cognitiveContentsafety 'Microsoft.CognitiveServices/accounts@2023-10-01
   location: location
   tags: tags
   sku: {
-    name: 'S0'
+    name: 'S0'  // Pay-as-you-go
   }
   kind: 'ContentSafety'
   properties: {
@@ -278,7 +281,7 @@ resource cognitiveOpenai 'Microsoft.CognitiveServices/accounts@2023-10-01-previe
   location: openaiLocation
   tags: tags
   sku: {
-    name: 'S0'
+    name: 'S0'  // Pay-as-you-go
   }
   kind: 'OpenAI'
   properties: {
@@ -291,7 +294,7 @@ resource gpt 'Microsoft.CognitiveServices/accounts/deployments@2023-10-01-previe
   name: gptModelFullName
   sku: {
     capacity: 50
-    name: 'Standard'
+    name: 'Standard'  // Pay-as-you-go
   }
   properties: {
     // raiPolicyName: contentfilter.name
@@ -367,7 +370,7 @@ resource ada 'Microsoft.CognitiveServices/accounts/deployments@2023-10-01-previe
   name: adaModelFullName
   sku: {
     capacity: 50
-    name: 'Standard'
+    name: 'Standard'  // Pay-as-you-go
   }
   properties: {
     // raiPolicyName: contentfilter.name
@@ -460,7 +463,7 @@ resource search 'Microsoft.Search/searchServices@2023-11-01' = {
   location: searchLocation
   tags: tags
   sku: {
-    name: 'basic'
+    name: 'basic'  // Smallest with semantic search
   }
   identity: {
     type: 'SystemAssigned'
@@ -469,3 +472,16 @@ resource search 'Microsoft.Search/searchServices@2023-11-01' = {
     semanticSearch: 'standard'
   }
 }
+
+resource translate 'Microsoft.CognitiveServices/accounts@2023-10-01-preview' = {
+  name: '${prefix}-translate'
+  location: location
+  tags: tags
+  sku: {
+    name: 'S1'  // Pay-as-you-go
+  }
+  kind: 'TextTranslation'
+  properties: {
+    customSubDomainName: '${prefix}-translate'
+  }
+}
diff --git a/bicep/main.bicep b/bicep/main.bicep
@@ -42,7 +42,6 @@ module app 'app.bicep' = {
     botCompany: botCompany
     botName: botName
     botPhoneNumber: botPhoneNumber
-    botVoiceName: 'fr-FR-DeniseNeural'
     gptModel: gptModel
     gptVersion: gptVersion
     imageVersion: imageVersion

diff --git a/helpers/config_models/ai_translation.py b/helpers/config_models/ai_translation.py
@@ -0,0 +1,7 @@
+from pydantic import SecretStr
+from pydantic_settings import BaseSettings
+
+
+class AiTranslationModel(BaseSettings):
+    access_key: SecretStr
+    endpoint: str
diff --git a/helpers/config_models/communication_service.py b/helpers/config_models/communication_service.py
@@ -11,4 +11,3 @@ class CommunicationServiceModel(BaseSettings):
     access_key: SecretStr
     endpoint: str
     phone_number: PhoneNumber
-    voice_name: str
diff --git a/helpers/config_models/database.py b/helpers/config_models/database.py
@@ -4,7 +4,7 @@
 from typing import Optional
 
 
-class Mode(str, Enum):
+class ModeEnum(str, Enum):
     COSMOS_DB = "cosmos_db"
     SQLITE = "sqlite"
 
@@ -32,17 +32,17 @@ def full_path(self) -> str:
 
 class DatabaseModel(BaseSettings):
     cosmos_db: Optional[CosmosDbModel] = None
-    mode: Mode = Mode.SQLITE
+    mode: ModeEnum = ModeEnum.SQLITE
     sqlite: Optional[SqliteModel] = None
 
     @validator("cosmos_db", always=True)
     def check_cosmos_db(cls, v, values, **kwargs):
-        if not v and values.get("mode", None) == Mode.COSMOS_DB:
+        if not v and values.get("mode", None) == ModeEnum.COSMOS_DB:
             raise ValueError("Cosmos DB config required")
         return v
 
     @validator("sqlite", always=True)
     def check_sqlite(cls, v, values, **kwargs):
-        if not v and values.get("mode", None) == Mode.SQLITE:
+        if not v and values.get("mode", None) == ModeEnum.SQLITE:
             raise ValueError("Sqlite config required")
         return v
diff --git a/helpers/config_models/monitoring.py b/helpers/config_models/monitoring.py
@@ -2,7 +2,7 @@
 from pydantic_settings import BaseSettings
 
 
-class LoggingLevel(str, Enum):
+class LoggingLevelEnum(str, Enum):
     # Copied from https://docs.python.org/3.12/library/logging.html#logging-levels
     CRITICAL = "CRITICAL"
     DEBUG = "DEBUG"
@@ -13,8 +13,8 @@ class LoggingLevel(str, Enum):
 
 
 class LoggingMonitoringModel(BaseSettings):
-    app_level: LoggingLevel = LoggingLevel.INFO
-    sys_level: LoggingLevel = LoggingLevel.WARNING
+    app_level: LoggingLevelEnum = LoggingLevelEnum.INFO
+    sys_level: LoggingLevelEnum = LoggingLevelEnum.WARNING
 
 
 class MonitoringModel(BaseSettings):