Convert the output from Google Cloud Speech Recognition to ELRC.

convert.py

"""
Convert the output from Google Cloud Speech Recognition to ELRC.
"""

import json, sys


class ELRC:
    def __init__(self):
        self.lines = []

    def add_line(self):
        if not self.lines or self.lines[-1]:
            self.lines.append([])

    def add_word(self, word):
        line = self.lines[-1]
        if len(line) >= 20:
            self.add_line()

        line.append(word)


def convert(response):
    """
    ELRC looks like this:

    [00:00.000] sdfasdf asdf <00:01.532> asd <00:01.884> fas <00:02.566> dfa sdf asdf asdf as <00:02.951> dfasdfa
    [00:03.180] asdf asdf asdf asdf
    """

    elrc = ELRC()

    response = response['response']    
    for result in response['results']:
        # Between each result there is usually a small pause (it seems!), so start a new line on each.
        elrc.add_line()

        main_alternative = result['alternatives'][0]

        for word in main_alternative['words']:
            elrc.add_word(word)


    # Format as text
    lines_text = []
    for line in elrc.lines:
        line_text = [f'<{word["startTime"][:-1]}> {word["word"]}' for word in line]

        line_text.insert(0, f'[{line[0]["startTime"][:-1]}]')
        lines_text.append(" ".join(line_text))

    return "\n".join(lines_text)




with open(sys.argv[1]) as f:
    content = json.loads(f.read())
    print(convert(content))