[Python] Creating a transcript with user timestamps for a Discord bot

I am using py-cord and the bot joins a meeting, starts recording, and once the end command is given it puts everything in this Discord sink.

At the moment, my code does transcription only relative to when a user starts speaking. Example: If a user starts speaking from the start of the meeting and ends 5 mins post meeting start, and at that point a 2nd user starts speaking - in the transcription the two will be as if they spoke at the same time. The transcript is not relative to the meeting start. Edit: which is what I am looking for. The transcript should follow the proper timeline from meeting start and when people started speaking relative to the meeting start. Example. Meeting starts at 0, Speaker 1 starts speaking until min 5. At min 5:01 Speaker 2 starts speaking, Meeting ends at min 10. The transcript should be something like

[00:00] Speaker 1:...
[05:01] Speaker 2:...


If they happen to speak at the same time at some point then both would appear in the transcript at that hr/min/second.

Appreciate any help.

async def once_done(sink: discord.sinks, channel: discord.TextChannel, *args):
meeting_start = connections.get(channel.guild.id, {}).get("meeting_start", 0)
user_start_times = connections.get(channel.guild.id, {}).get("user_start", {})

await channel.send(" Transcribing...")
await sink.vc.disconnect()

words_list = []

for user_id, audio in sink.audio_data.items():
payload = {"buffer": audio.file.read()}
response = deepgram.listen.rest.v("1").transcribe_file(payload, options)

words = response["results"]["channels"][0]["alternatives"][0]["words"]
words = [word.to_dict() for word in words]

if words:
first_word_time = words[0]["start"]
if user_id not in user_start_times:
user_start_times[user_id] = first_word_time

user_offset = user_start_times[user_id] - meeting_start

for word in words:
words_list.append({
"word": word["word"],
"start": word["start"] + user_offset,
"end": word["end"] + user_offset,
"punctuated_word": word["punctuated_word"],
"speaker": user_id,
})

words_list.sort(key=lambda x: x["start"])

transcript = []
current_speaker = None
current_segment = {"speaker": None, "start": None, "end": None, "text": ""}

for word in words_list:
speaker = word["speaker"]
if (speaker != current_speaker or
(current_segment["end"] is not None and word["start"] - current_segment["end"] > 2)):
if current_speaker is not None:
transcript.append(current_segment)
current_speaker = speaker
current_segment = {
"speaker": speaker,
"start": word["start"],
"end": word["end"],
"text": word["punctuated_word"]
}
else:
current_segment["text"] += " " + word["punctuated_word"]
current_segment["end"] = word["end"]

if current_segment["speaker"] is not None:
transcript.append(current_segment)

with open("transcript.txt", "w") as f:
for segment in transcript:
start_minutes = int(segment["start"] // 60)
start_seconds = int(segment["start"] % 60)
end_minutes = int(segment["end"] // 60)
end_seconds = int(segment["end"] % 60)
f.write(f"({start_minutes}:{start_seconds:02d} - {end_minutes}:{end_seconds:02d}) {segment['speaker']}:\n{segment['text']}\n\n")

await channel.send(file=discord.File("transcript.txt"))
await channel.send(" Transcription finished.")

Continue reading...