How to solve error when using vosk library "Ivector feature dimension mismatch: got -1 but network expects 40"?

Question

Python

I did speech recognition using the vosk library, but it gives an error.

Here is part of the code:

import json
import pyaudio
from vosk import Model, KaldiRecognizer

p=pyaudio.PyAudio()
model = Model('vosk-model-small-ru-0.15')
rec = KaldiRecognizer(model, 8000)
streaming = p.open(format=pyaudio.paInt16, channels=1,rate = 8000, input=True, frames_per_buffer=8000)
streaming.start_stream()
engine = pyttsx3.init()

Here is the error:


ERROR (VoskAPI:DecodableNnetLoopedOnlineBase():decodable-online-looped.cc:50) Ivector feature dimension mismatch: got -1 but network expects 40
terminate called after throwing an instance of 'kaldi::KaldiFatalError'
what(): kaldi::KaldiFatalError


As I understand it, the problem is here:

Ivector feature dimension mismatch: got -1 but network expects 40

How to solve it? I have no idea

Answer 1

There is no ivector directory in the model, it should be there. The model may have been corrupted while downloading. Need to download again.