feat: Add support for Russian and Kazakh languages in whisper model

.env.sample

@@ -15,4 +15,6 @@ TWILIO_PHONE_NUMBER=
 # Plivo credentials
 PLIVO_AUTH_ID=
 PLIVO_AUTH_TOKEN=
-PLIVO_PHONE_NUMBER=
+PLIVO_PHONE_NUMBER=
+
+NGROK_AUTHTOKEN=

.gitignore

@@ -167,4 +167,6 @@ agent_data/**/mp3
 */__pycache__/
 */*/__pycache__/
 logs/
-agent_data/
+agent_data/
+
+.env

bolna/models.py

@@ -80,17 +80,22 @@ class Transcriber(BaseModel):
     encoding: Optional[str] = "linear16"
     endpointing: Optional[int] = 400
     keywords: Optional[str] = None
-    task:Optional[str] = "transcribe"
+    task: Optional[str] = "transcribe"
     provider: Optional[str] = "deepgram"
 
     @validator("provider")
     def validate_model(cls, value):
         print(f"value {value}, PROVIDERS {list(SUPPORTED_TRANSCRIBER_PROVIDERS.keys())}")
         return validate_attribute(value, list(SUPPORTED_TRANSCRIBER_PROVIDERS.keys()))
 
+    # Only whisper works well with russian and kazakh
     @validator("language")
-    def validate_language(cls, value):
-        return validate_attribute(value, ["en", "hi", "es", "fr", "pt", "ko", "ja", "zh", "de", "it", "pt-BR"])
+    def validate_language(cls, value, values):
+        supported_languages = ["en", "hi", "es", "fr", "pt", "ko", "ja", "zh", "de", "it", "pt-BR"]
+        if values.get('model') == 'whisper':
+            supported_languages.append('ru')
+            supported_languages.append('kk')
+        return validate_attribute(value, supported_languages)
 
 
 class Synthesizer(BaseModel):

local_setup/docker-compose.yml

@@ -36,6 +36,8 @@ services:
       - ./ngrok-config.yml:/etc/ngrok.yml
     ports:
       - 4040:4040
+    environment:
+      - NGROK_AUTHTOKEN=${NGROK_AUTHTOKEN}
 
   ### Telephony servers ###
   twilio-app:

local_setup/ngrok-config.yml

@@ -1,6 +1,5 @@
 region: us
 version: '2'
-authtoken: <your-auth-token>
 tunnels:
   twilio-app:
     addr: twilio-app:8001