Error on Stat: 2

[ankushnarula]

The documentation isn't very clear about M1 support for native client. But in any case, I tried the following

% ./stt --model model.tflite --scorer huge-vocabulary.scorer -audio test.wav

and received:

TensorFlow: v2.8.0-8-g06c8fea58fd
 Coqui STT: v1.4.0-alpha.1-0-gf3cdbd77
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
Error on stat: 2
Unexpected type for udio: 0

Having never used STT or TensorFlow, I have no idea where to begin. Any pointers?

[wasertech]

@ankushnarula Can you show us the output of soxi for test.wav? It seem you have an unexpected type of audio.
It should look something like this:

❯ soxi ~/.assistant/data/fr/SR/savewav_2022-04-11_17-18-22_828005.wav

Input File     : '/home/waser/.assistant/data/fr/SR/savewav_2022-04-11_17-18-22_828005.wav'
Channels       : 1
Sample Rate    : 16000
Precision      : 16-bit
Duration       : 00:00:07.68 = 122880 samples ~ 576 CDDA sectors
File Size      : 246k
Bit Rate       : 256k
Sample Encoding: 16-bit Signed Integer PCM

Make sure your audio validates the following requirements:

1. Channels : 1
2. Sample Rate : 16000
3. Precision : 16-bit
4. Bit Rate : 256k
5. Sample Encoding: 16-bit Signed Integer PCM