## process wav, split to 8 min parts and produce output.txt
import speech_recognition as sr
from pydub import AudioSegment
import math
import datetime

def format_timestamp(milliseconds):
    seconds = milliseconds / 1000
    return str(datetime.timedelta(seconds=seconds)).split('.')[0]

def split_and_recognize(wav_file, chunk_length_ms=480000):  # 480000 ms = 8 minutes 0 seconds
    audio = AudioSegment.from_wav(wav_file)
    total_duration = len(audio)
    chunks = math.ceil(total_duration / chunk_length_ms)

    recognizer = sr.Recognizer()
    full_text = []

    for i in range(chunks):
        start_time = i * chunk_length_ms
        end_time = min((i + 1) * chunk_length_ms, total_duration)

        chunk = audio[start_time:end_time]
        chunk_file = f"temp_chunk_{i}.wav"
        chunk.export(chunk_file, format="wav")

        with sr.AudioFile(chunk_file) as source:
            audio_data = recognizer.record(source)

        try:
            print(f"Recognizing chunk {i + 1} of {chunks}...")
            text = recognizer.recognize_google(audio_data, language="sl-SI")
            timestamp = format_timestamp(start_time)
            full_text.append(f"[{timestamp}] {text}")
        except sr.UnknownValueError:
            print(f"Chunk {i + 1}: Speech recognition could not understand the audio")
        except sr.RequestError as e:
            print(f"Chunk {i + 1}: Could not request results; {e}")

        # Clean up temporary file
        import os
        os.remove(chunk_file)

    return "\n\n".join(full_text)  # Join with double newline for paragraph separation

# Usage
wav_file = "r:\\sandi.wav"
result = split_and_recognize(wav_file)

# Write result to file
with open("r:\\output.txt", "w", encoding="utf-8") as output_file:
    output_file.write(result)

print("Transcription complete. Text written to r:\\output.txt")