Timestamps while working with code

[hussamsayeed]

How do I get the timestamps while working with the code? I see timestamps are visible while working with command.

[levaweb]

I've seen in the examples the keyword: without_timestamps=True/False
Can we use it to get the timestamp printed out when executing the code in Python?
Or, how do we get the timestamp in a separate file?
It looks weird that timestamps are only showing up in the CMD prompt mode.
Thanks

[jongwook]

They are returned in the field result["segments"].

[hussamsayeed]

since the output is in dictionary format, we have to open it as iterator, you can add this code to get the time stamps.

import whisper
import zlib
from typing import Iterator, TextIO

def format_timestamp(seconds: float, always_include_hours: bool = False, decimal_marker: str = '.'):
    assert seconds >= 0, "non-negative timestamp expected"
    milliseconds = round(seconds * 1000.0)

    hours = milliseconds // 3_600_000
    milliseconds -= hours * 3_600_000

    minutes = milliseconds // 60_000
    milliseconds -= minutes * 60_000

    seconds = milliseconds // 1_000
    milliseconds -= seconds * 1_000

    hours_marker = f"{hours:02d}:" if always_include_hours or hours > 0 else ""
    return f"{hours_marker}{minutes:02d}:{seconds:02d}{decimal_marker}{milliseconds:03d}"

def transcript_timestamp(transcript: Iterator[dict]):
    for segment in transcript:
        print(
            f"{format_timestamp(segment['start'])} --> {format_timestamp(segment['end'])} "
            f"Text: {segment['text'].strip().replace('-->', '->')}\n",
            flush=True
        )

Call the function by feeding the output

trans=transcript_timestamp(result["segments"])

[levaweb]

Thanks Hussam! Really appreciated

Is there a way to improve the accuracy of the timestamps?
On average they looks +/-250 ms inaccurate.
Is this perhaps something due to rounding floats into integers?

Thanks

[hussamsayeed]

Check with this it may works #435

[levaweb]

Awesome! Thanks for sharing

[tflax32]

Here is a way to get the srt file from your whisper transcription using the pysrt library:

import whisper
import pysrt

model = whisper.load_model("base")
audio_file = ("YOUR AUDIO FILE PATH")
result = model.transcribe(audio_file, language='en')

subs = pysrt.SubRipFile()
sub_idx = 1

for i in range(len(result["segments"])):
    start_time = result["segments"][i]["start"]
    end_time = result["segments"][i]["end"]
    duration = end_time - start_time
    timestamp = f"{start_time:.3f} - {end_time:.3f}"
    text = result["segments"][i]["text"]
    
    sub = pysrt.SubRipItem(index=sub_idx, start=pysrt.SubRipTime(seconds=start_time), 
                           end=pysrt.SubRipTime(seconds=end_time), text=text)
    subs.append(sub)
    sub_idx += 1
    
subs.save("output.srt")