[Python] Twilio TTS Mispronounces "Aradhya" – How to Fix It?

I’m using Twilio's text-to-speech (TTS) in a Django view to read out a message before connecting a call. However, Twilio mispronounces the name "Aradhya", and I’m looking for a way to fix it.

Problem:


Twilio reads "Aradhya" incorrectly, and I want it to sound like "Uh-raa-dhyaa" (अाराध्य in Hindi).

Code (Simplified Version):


from django.http import HttpResponse
from django.views.decorators.csrf import csrf_exempt
from twilio.twiml.voice_response import VoiceResponse

@csrf_exempt
def connect_call(request):
response = VoiceResponse()
customer_name = "John Doe" # Example; this is fetched dynamically in actual code
response.say(f"Hello {customer_name}, I am Aradhya. Please hold while we connect you.")
response.dial("+919836731154") # Connect call

return HttpResponse(str(response), content_type="application/xml")

What I Tried:

1. Changing the spelling (e.g., "Aaraadhyaa", "Aradyaa") → No improvement.
2. Using SSML <phoneme> tag:
   <Response>
   <Say>
   I am <phoneme alphabet="ipa" ph="əˈrɑːðjə">Aradhya</phoneme>.
   </Say>
   </Response>
   
   Issue: Twilio still mispronounces it.
3. Adding pauses with <break>:
   <Say>Hello, I am <break time="200ms"/> Aaraadhyaa.</Say>
   
   Issue: Slight improvement, but not perfect.
4. Using Twilio Neural Voices (Amazon Polly):
   <Response>
   <Say voice="Polly.Aditi">Hello, I am Aradhya.</Say>
   </Response>
   
   Issue: Not all Twilio accounts support Polly voices.

Question:


How can I get Twilio TTS to pronounce "Aradhya" correctly without using pre-recorded audio?

Any help would be appreciated!

Continue reading...