Quick (and dirty) Python script that takes the transcript information from CSPAN video pages and parses it into a JSON file with the time stamp, speaker name, and speech text. There are probably 100 more elegant ways to write this, but it works for what I needed.

CSPAN-transcript-parser.py

from html.parser import HTMLParser
import re
import json
import datetime as dt

class Entry():
    def __init__(self, time):
        self.time = ''
        self.time_base = time
        self.speaker = ''
        self.text = ''

    @property
    def value(self):
        return {
            'time': self.time,
            'time_utc': self.get_utc_time(self.time_base, self.time),
            'speaker': self.speaker,
            'text': re.sub('(\.){2,}$','.',self.text)
            }

    @staticmethod
    def get_utc_time(base, delta):
        hr, min, sec = map(float, delta.split(':'))
        td = dt.timedelta(hours=hr, minutes=min, seconds=sec)
        return (base + td).isoformat() + 'Z'
        
        

class HTMLParseTable(HTMLParser):
    __outer_class = 'table-wrap load-transcript'
    __outer_tag = 'div'
    __inner_tags = {
            'th': 0,
            'strong': 0,
            'p': 0
        }
    
    __outer_capture = False
    __outer_instances = 0

    __entry = None
    __output = []
    __time = ''

    @property
    def output(self):
        return self.__output

    @property
    def time_start(self):
        return self.__time
    
    @time_start.setter
    def time_start(self, value):
        self.__time = dt.datetime.strptime(value, "%Y-%m-%dT%H:%M:%SZ" )

    
    def handle_starttag(self, tag, attrs):
        if not self.__outer_capture:        
            if tag == self.__outer_tag:
                if self.has_attribute(attrs, 'class', self.__outer_class):
                    self.__outer_capture = True
                    print('Found outer tag!')
                    
                if self.__outer_capture:
                    self.__outer_instances += 1
        if self.__outer_capture:
            if tag in self.__inner_tags.keys():
                self.__inner_tags[tag] += 1
            
    def handle_endtag(self, tag):
        if self.__outer_capture:
            if tag == self.__outer_tag:
                self.__outer_instances -= 1
            elif tag in self.__inner_tags.keys():
                self.__inner_tags[tag] -= 1
                if tag == 'p':
                    self.__output.append(self.__entry.value)

            if self.__outer_instances == 0:
                self.__outer_capture = False
                print('Found end tag!')

    def handle_data(self, data):
        if self.__outer_capture:
            if self.__inner_tags.get('th') > 0:
                m = re.search('(\d{2}\:\d{2}\:\d{2})', data)
                if m:
                    self.__entry = Entry(self.time_start)
                    self.__entry.time = m.group(1)
            elif self.__inner_tags.get('strong') > 0:
                self.__entry.speaker = data
            elif self.__inner_tags.get('p') > 0:
                self.__entry.text += data


    @staticmethod
    def has_attribute(attrs, key, value):
        found = False
        for k,v in attrs:
            if k == key and v == value:
                found = True
        return found
            
# Set the start time of the event as YYYY-MM-DDTHH:MM:SSZ
parser = HTMLParseTable()
ts = input('Event UTC Start Time (YYYY-MM-DDTHH:MM:SSZ):')
parser.time_start = ts

with open('C:/yourdir/input.html','r') as f:
    s = f.read()
    parser.feed(s)

with open('C:/yourdir/output.json', 'w') as f:
    json.dump(parser.output, f, sort_keys=True,indent=4, separators=(',', ': '))

Can be tested with https://www.c-span.org/video/?414225-1/presidential-nominees-debate-hofstra-university&start=5539

Just noticed that some parts of the CSPAN transcript are repeated word-for-word. These are clearly mistakes as they don't even sync to the video. You should parse and select the 1st instance of when something was said as the correct time

Any chance of the cleaned up data?