Hello,
I’m working with an M5Stack AX630C device and I’m trying to get Automatic Speech Recognition (ASR) working with the Sherpa NCNN Streaming Zipformer model. However, I’ve encountered an issue where the model fails to load, and I keep receiving an error message: "Model loading failed."
Here are the details of the setup and error I am encountering:
Setup Details:
Device: M5Stack AX630C (connected to Raspberry Pi running Ubuntu. I've connected to the AX630C via ADB shell.
I’ve been using the following Python script to configure the ASR model and push the setup command and when I execute it I get the following message:
Setup ASR Response: {"created":1692663788,"data":"None","error":{"code":-5,"message":"Model loading failed."},"object":"None","request_id":"21","work_id":"asr"}
Has anyone encountered a similar issue with loading ASR models on the M5Stack AX630C? What steps did you take to resolve it?
Are there any additional steps or configuration changes that I might be missing to ensure the model loads correctly?
Could this be related to firmware versions or dependencies that are not properly set up for ASR?
Any insights or suggestions would be greatly appreciated!
SCRIPT I AM RUNNING
import socket
# Connect to localhost:10001
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.connect(("127.0.0.1", 10001))
# Corrected ASR setup message
message = """
{
"request_id": "21",
"work_id": "asr",
"action": "setup",
"object": "asr.setup",
"data": {
"model": "mode_sherpa-ncnn-streaming-zipformer-20M-2023-02-17.json",
"response_format": "asr.text",
"input": ["sys.pcm"]
}
}
"""
# Send the setup message
s.sendall((message.strip() + "\n").encode())
# Receive and print response
response = s.recv(4096)
print("Setup ASR Response:", response.decode())
s.close()
