Hello guys, when I am calling my assistant through Vapi website, it is working perfectly, but when I connect it to a phone number (telnyx), it doesn't recognize my voice, or even don't answer to my sentences. It seems to me that doesn't recognize my voice. I tried every transcriber and every transcriber performed badly. What can be the issue guys? Should I tweak something in the Provider Keys tab? Let me know please. I am doing hungarian language. Thank you @User

Hi there! 👋 I understand you're having trouble with voice recognition when using your Telnyx phone number connection, even though it works fine on the Vapi website. I can't provide exact settings to tweak specifically for Hungarian with Telnyx, but I can definitely help with several aspects of your issue based on information I've found. ## Possible Causes & Solutions ### Language Configuration Make sure your transcriber is correctly set to Hungarian. According to [Azure Speech Transcriber documentation](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1341522069946499093), you can use the 'hu-HU' language code for Hungarian. For other transcribers, you'd set the language parameter in your configuration:

"transcriber": {
  "model": "your-model",
  "language": "hu",
  "provider": "your-provider"
}

### Transcriber Settings for Non-English Languages For Hungarian, consider these adjustments: 1. **Disable Smart Endpointing**: The

smartEndpointingEnabled

setting is optimized for English and should be set to

false

when using Hungarian, [according to this source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1329414710583627796). 2. **Keep Chunk Plan Enabled**: For non-English languages, it's important to keep

chunkPlan.enabled: true

for proper sentence boundary detection and improved speech flow, [as mentioned here](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1343543475383046185). 3. **Adjust Endpointing Settings**: Try setting

startSpeakingPlan.waitSeconds

to 0.8 and

transcriber.endpointing

to 300 for better results, [as suggested here](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1336875140990111796). ### Telnyx-Specific Issues

Hey! To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help us figure out what went wrong!

I tried plenty of times, very inaccurate a lot of times, 18e3c2ae-b331-4026-a835-ce41c94f1ba0 this call was yesterday at 16:33 UTC Didn't understand me, I am hungarian It doesn't say anything, or says something that isn't related to what I said

Hey Enyedi Péter, You have to be more specific about what response you actually got compared to what you were expecting. This is quite generic, so I couldn't understand what exactly went wrong. I can see that the output is generated in Hungarian, but I cannot understand what went wrong. So I hope you could understand why I require the information in the following order and what exact information I am asking for.