MeloTTS fails to convert text to speech with Verbi

[skyy5skyy6]

Hi
I am trying to run MeloTTS locally on Raspberry Pi 5 with Verbi.
The MeloTTS endpoint starts correctly but fails when text is to be converted to audio

2025-07-28 20:02:33,053 - INFO - Calibrating for ambient noise...
2025-07-28 20:02:34,073 - INFO - Recording started
2025-07-28 20:02:38,905 - INFO - Recording complete
2025-07-28 20:02:41,965 - INFO - You said: It's about to get in.
2025-07-28 20:02:47,415 - INFO - HTTP Request: POST http://127.0.0.1:11434/api/chat "HTTP/1.1 200 OK"
2025-07-28 20:02:47,416 - INFO - Response: Sounds like someone is arriving soon! What can I do for you?
2025-07-28 20:02:47,436 - ERROR - Failed to convert text to speech: 500 Server Error: Internal Server Error for url: http://localhost:5150/generate-audio/
2025-07-28 20:02:47,447 - ERROR - Failed to play audio: No file 'output.mp3' found in working directory '/home/sankum/Verbi'.

The MeloTTS API endpoint started correctly but failed"

WeightNorm.apply(module, name, dim)
INFO: Started server process [2378]
INFO: Waiting for application startup.
INFO: Application startup complete.
INFO: Uvicorn running on http://0.0.0.0:5150 (Press CTRL+C to quit)

Text split to sentences.
Sounds like someone is arriving soon! What can I do for you?

0%| | 0/1 [00:00<?, ?it/s]
INFO: 127.0.0.1:56080 - "POST /generate-audio/ HTTP/1.1" 500 Internal Serve

Can somebody point to a solution for this? Thanks.