from stt import Model
import gradio as gr
import numpy as np

model = 'stt-comodoro-czech-2022-05-31.tflite'
scorer = 'czech-large-vocab.scorer'
beam_width = 512
lm_alpha = 0.94
lm_beta = 2.52

model = Model(model)
model.enableExternalScorer(scorer)
model.setScorerAlphaBeta(lm_alpha, lm_beta)
model.setBeamWidth(beam_width)

def reformat_freq(sr, y):
    if sr not in (
        48000,
        16000,
    ):  # Deepspeech only supports 16k, (we convert 48k -> 16k)
        raise ValueError("Unsupported rate", sr)
    if sr == 48000:
        y = (
            ((y / max(np.max(y), 1)) * 32767)
            .reshape((-1, 3))
            .mean(axis=1)
            .astype("int16")
        )
        sr = 16000
    return sr, y

def transcribe(speech):
    _, y = reformat_freq(*speech)
    stream = model.createStream()
    stream.feedAudioContent(y)
    text = stream.intermediateDecode()
    return text

with gr.Blocks() as blocks:
  audio = gr.Audio(source="microphone", type="numpy", streaming=False,
      label='Pokud je to třeba, povolte mikrofon pro tuto stránku, \
 klikněte na Record from microphone, po dokončení nahrávání na Stop recording a poté na Rozpoznat')
  btn = gr.Button('Rozpoznat')
  output = gr.Textbox(show_label=False)
  btn.click(fn=transcribe, inputs=[audio], 
      outputs=[output])

blocks.launch(enable_queue=True, debug=True, share=True)