@chia5966
I'm sorry to hear about the inconsistencies you've been experiencing with voice speed and quality. Let's address these issues one by one to ensure a smoother experience for you.
For PlayHT and ensuring consistent pacing when reading phone numbers, you can try adding strategic pauses in the text you're sending to the TTS engine. For example, you could format the number with pauses like this: "6 1 7 ... 5 5 5 ... 1 2 1 2". The ellipsis (...) represents a brief pause, which might help with pacing.
Regarding the speech speed variations and unexpected noises, it might be helpful to check if there are any settings within PlayHT that allow you to control the speech rate and quality more precisely. If the issues persist, reaching out to PlayHT support with specific examples could help them troubleshoot the problem.
For collecting email addresses, it's a common challenge due to the complexity and variability of email formats. One approach is to use voice prompts to guide the user to speak their email address slowly and clearly, and possibly spell out parts that are commonly misunderstood.
If ElevenLabs was working well for you before, it might be worth reaching out to their support to see if they can provide any stability updates or workarounds.
In the meantime, if you'd like, I can assist you with configuring your voice API settings to try and mitigate these issues. Would you like to proceed with that?