Spaces:

sanchit-gandhi

/

whisper-jax

like 2.33k

Running

did you figure it out?

here's the python script to do this if anyone who needs it. Just change the input_file location:

import re

# Function to convert timestamp to seconds
def time_to_seconds(timestamp):
    m, s = map(float, timestamp.split(':'))
    return m * 60 + s

# Function to format time in seconds as SRT time (hh:mm:ss,ms)
def format_time(seconds):
    milliseconds = int((seconds - int(seconds)) * 1000)
    seconds = int(seconds)
    minutes, seconds = divmod(seconds, 60)
    hours, minutes = divmod(minutes, 60)
    return f"{hours:02d}:{minutes:02d}:{seconds:02d},{milliseconds:03d}"

# Input and output file names
input_file = r"C:\Documents\Python scripts\input.txt"
output_file = r"C:\Documents\Python scripts\output.srt"

# Open the input and output files
with open(input_file, "r") as infile, open(output_file, "w") as outfile:
    # Read lines from the input file
    lines = infile.readlines()
    
    # Initialize subtitle index
    subtitle_index = 1
    
    # Iterate through lines and convert to SRT format
    for line in lines:
        # Use regular expression to extract start and end times
        time_match = re.search(r'\[(\d+:\d+\.\d+) -> (\d+:\d+\.\d+)]', line)
        
        if time_match:
            start_time = time_to_seconds(time_match.group(1))
            end_time = time_to_seconds(time_match.group(2))
            outfile.write(f"{subtitle_index}\n")
            outfile.write(f"{format_time(start_time)} --> {format_time(end_time)}\n")
            subtitle_index += 1
        else:
            # Skip lines without time information
            continue
        
        # Write the subtitle text
        text = re.sub(r'\[.*\]', '', line).strip()
        outfile.write(f"{text}\n\n")

print(f"Conversion completed. The output has been saved to {output_file}")

const fs = require('fs');

// Funkcja przekształcająca napisy w formacie "whisper" na format SRT
function convertWhisperToSrt(data) {
const lines = data.split('\n');
let srtOutput = '';
let subtitleIndex = 1;

for (const line of lines) {
    if (line.trim() !== '') {
        const match = line.match(/\[(\d+:\d+\.\d+) --> (\d+:\d+\.\d+)](.*)/);
        if (match && match.length === 4) {
            const startTime = formatTime(match[1]);
            const endTime = formatTime(match[2]);
            const text = match[3].trim();

            srtOutput += `${subtitleIndex}\n${startTime} --> ${endTime}\n${text}\n\n`;
            subtitleIndex++;
        }
    }
}

return srtOutput;

}

// Funkcja do formatowania czasu w stylu "00:00:00,000 --> 00:00:01,960"
function formatTime(timestamp) {
const parts = timestamp.split(':');
const minutes = parts[0];
const secondsAndMillis = parts[1].split('.');
const seconds = secondsAndMillis[0];
const milliseconds = secondsAndMillis[1].padEnd(3, '0');

return `00:${pad(minutes)}:${pad(seconds)},${milliseconds}`;

}

// Funkcja do wypełniania zerami
function pad(number, length = 2) {
return number.toString().padStart(length, '0');
}

// Odczyt pliku "whisper.txt" i przetwarzanie danych
const inputFile = "whisper.txt";
try {
const data = fs.readFileSync(inputFile, 'utf8');
const srtOutput = convertWhisperToSrt(data);

// Zapis przekonwertowanych napisów do pliku "converted.srt"
const outputFileName = "german2.srt";
fs.writeFileSync(outputFileName, srtOutput, 'utf8');

console.log(`Plik "${outputFileName}" został zapisany w formacie SRT.`);

} catch (err) {
console.error(Błąd podczas odczytu/zapisu plików: ${err.message});
}
here is js node script that convert a file txt to srt.Remember to change "-" to "--" in txt file after that it going to work easyly.

I just copy and paste the output into a text file, and then I load that text file into the free Subtitle Edit program - https://www.nikse.dk/subtitleedit - Then I run "Fix common errors" to have it auto-fix any issues and then export it as an SRT.

I adjust CuddleMaster' code above to match both format (Hours:Minutes:Seconds.Milliseconds) (01:03:52.328) or (Minutes:Seconds.Milliseconds)(21:32.545);
and remove the time convert

import re

# Input and output file names
input_file = r"./input.txt"
output_file = r"./output.srt"

# Open the input and output files
with open(input_file, "r") as infile, open(output_file, "w") as outfile:
    # Read lines from the input file
    lines = infile.readlines()
    
    # Initialize subtitle index
    subtitle_index = 1
    
    # Iterate through lines and convert to SRT format
    for line in lines:
        # Use regular expression to extract start and end times, both format (Hours:Minutes:Seconds.Milliseconds) (01:03:52.328) or (Minutes:Seconds.Milliseconds)(21:32.545)
        time_match = re.search(r'\[((\d+:){1,2}\d+\.\d+) -> ((\d+:){1,2}\d+\.\d+)]', line)
        
        if time_match:
            start_time = time_match.group(1).replace(".", ",")
            if len(start_time) < 10:    # if time string like 34:08,342, lengs <= 9
                start_time = "00:" + start_time     # add hour '00:'
            end_time = time_match.group(3).replace(".", ",")
            if len(end_time) < 10:
                end_time = "00:" + end_time
            outfile.write(f"{subtitle_index}\n")
            outfile.write(f"{start_time} --> {end_time}\n")

            subtitle_index += 1
        else:
            # Skip lines without time information
            continue
        
        # Write the subtitle text
        text = re.sub(r'\[.*\]', '', line).strip()
        outfile.write(f"{text}\n\n")

print(f"Conversion completed. The output has been saved to {output_file}")

here's the python script to do this if anyone who needs it. Just change the input_file location:

import re

# Function to convert timestamp to seconds
def time_to_seconds(timestamp):
    m, s = map(float, timestamp.split(':'))
    return m * 60 + s

# Function to format time in seconds as SRT time (hh:mm:ss,ms)
def format_time(seconds):
    milliseconds = int((seconds - int(seconds)) * 1000)
    seconds = int(seconds)
    minutes, seconds = divmod(seconds, 60)
    hours, minutes = divmod(minutes, 60)
    return f"{hours:02d}:{minutes:02d}:{seconds:02d},{milliseconds:03d}"

# Input and output file names
input_file = r"C:\Documents\Python scripts\input.txt"
output_file = r"C:\Documents\Python scripts\output.srt"

# Open the input and output files
with open(input_file, "r") as infile, open(output_file, "w") as outfile:
    # Read lines from the input file
    lines = infile.readlines()
    
    # Initialize subtitle index
    subtitle_index = 1
    
    # Iterate through lines and convert to SRT format
    for line in lines:
        # Use regular expression to extract start and end times
        time_match = re.search(r'\[(\d+:\d+\.\d+) -> (\d+:\d+\.\d+)]', line)
        
        if time_match:
            start_time = time_to_seconds(time_match.group(1))
            end_time = time_to_seconds(time_match.group(2))
            outfile.write(f"{subtitle_index}\n")
            outfile.write(f"{format_time(start_time)} --> {format_time(end_time)}\n")
            subtitle_index += 1
        else:
            # Skip lines without time information
            continue
        
        # Write the subtitle text
        text = re.sub(r'\[.*\]', '', line).strip()
        outfile.write(f"{text}\n\n")

print(f"Conversion completed. The output has been saved to {output_file}")

Good lookin' out!

I adjust CuddleMaster' code above to match both format (Hours:Minutes:Seconds.Milliseconds) (01:03:52.328) or (Minutes:Seconds.Milliseconds)(21:32.545);
and remove the time convert

import re

# Input and output file names
input_file = r"./input.txt"
output_file = r"./output.srt"

# Open the input and output files
with open(input_file, "r") as infile, open(output_file, "w") as outfile:
    # Read lines from the input file
    lines = infile.readlines()
    
    # Initialize subtitle index
    subtitle_index = 1
    
    # Iterate through lines and convert to SRT format
    for line in lines:
        # Use regular expression to extract start and end times, both format (Hours:Minutes:Seconds.Milliseconds) (01:03:52.328) or (Minutes:Seconds.Milliseconds)(21:32.545)
        time_match = re.search(r'\[((\d+:){1,2}\d+\.\d+) -> ((\d+:){1,2}\d+\.\d+)]', line)
        
        if time_match:
            start_time = time_match.group(1).replace(".", ",")
            if len(start_time) < 10:    # if time string like 34:08,342, lengs <= 9
                start_time = "00:" + start_time     # add hour '00:'
            end_time = time_match.group(3).replace(".", ",")
            if len(end_time) < 10:
                end_time = "00:" + end_time
            outfile.write(f"{subtitle_index}\n")
            outfile.write(f"{start_time} --> {end_time}\n")

            subtitle_index += 1
        else:
            # Skip lines without time information
            continue
        
        # Write the subtitle text
        text = re.sub(r'\[.*\]', '', line).strip()
        outfile.write(f"{text}\n\n")

print(f"Conversion completed. The output has been saved to {output_file}")

Nice, thanks