Speech to Text

1
from openai import OpenAI
2
3
client = OpenAI(
4
    base_url="https://api.neura-ai.app/v1"
5
)
6
7
with open("audio.mp3", "rb") as audio_file:
8
    transcription = client.audio.transcriptions.create(
9
        model="whisper-1",
10
        file=audio_file
11
    )
12
13
print(transcription.text)

1
with open("spanish_audio.mp3", "rb") as audio_file:
2
    transcription = client.audio.transcriptions.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        language="es"  # ISO-639-1 code
6
    )
7
8
print(transcription.text)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    response_format="text"
5
)
6
7
print(transcription.text)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    response_format="json"
5
)
6
7
print(transcription.text)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    response_format="verbose_json"
5
)
6
7
print(f"Language: {transcription.language}")
8
print(f"Duration: {transcription.duration}")
9
10
for segment in transcription.segments:
11
    print(f"[{segment.start:.2f}s - {segment.end:.2f}s]: {segment.text}")

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    response_format="srt"
5
)
6
7
with open("subtitles.srt", "w") as f:
8
    f.write(transcription.text)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    response_format="vtt"
5
)
6
7
with open("subtitles.vtt", "w") as f:
8
    f.write(transcription.text)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    prompt="This is a technical discussion about machine learning and neural networks."
5
)

1
transcription = client.audio.transcriptions.create(
2
    model="whisper-1",
3
    file=audio_file,
4
    temperature=0.2  # Lower = more consistent, Higher = more varied
5
)

1
from openai import OpenAI
2
3
client = OpenAI(base_url="https://api.neura-ai.app/v1")
4
5
def transcribe_meeting(audio_path):
6
    with open(audio_path, "rb") as audio_file:
7
        transcription = client.audio.transcriptions.create(
8
            model="whisper-1",
9
            file=audio_file,
10
            response_format="verbose_json",
11
            prompt="Business meeting discussing Q4 targets and marketing strategy"
12
        )
13
    
14
    # Save full transcript
15
    with open("meeting_transcript.txt", "w") as f:
16
        f.write(transcription.text)
17
    
18
    # Save timestamped version
19
    with open("meeting_detailed.txt", "w") as f:
20
        for segment in transcription.segments:
21
            f.write(f"[{segment.start:.2f}s]: {segment.text}\n")
22
    
23
    return transcription
24
25
result = transcribe_meeting("quarterly_meeting.mp3")
26
print(f"Transcribed {result.duration:.2f} seconds of audio")

1
def transcribe_podcast(episode_file, episode_title):
2
    with open(episode_file, "rb") as audio_file:
3
        transcription = client.audio.transcriptions.create(
4
            model="whisper-1",
5
            file=audio_file,
6
            response_format="srt",
7
            prompt=f"Podcast episode: {episode_title}"
8
        )
9
    
10
    # Save as subtitle file
11
    srt_filename = episode_file.replace(".mp3", ".srt")
12
    with open(srt_filename, "w", encoding="utf-8") as f:
13
        f.write(transcription.text)
14
    
15
    print(f"Subtitles saved to {srt_filename}")
16
17
transcribe_podcast("episode_42.mp3", "The Future of AI")

1
def transcribe_interview(audio_path, interviewer, interviewee):
2
    with open(audio_path, "rb") as audio_file:
3
        transcription = client.audio.transcriptions.create(
4
            model="whisper-1",
5
            file=audio_file,
6
            response_format="verbose_json",
7
            prompt=f"Interview between {interviewer} and {interviewee}"
8
        )
9
    
10
    # Format with timestamps
11
    formatted_text = f"Interview: {interviewee}\n"
12
    formatted_text += f"Date: {transcription.duration:.0f} seconds\n\n"
13
    
14
    for segment in transcription.segments:
15
        timestamp = f"[{int(segment.start//60):02d}:{int(segment.start%60):02d}]"
16
        formatted_text += f"{timestamp} {segment.text}\n"
17
    
18
    return formatted_text
19
20
result = transcribe_interview(
21
    "interview.wav",
22
    "John Smith",
23
    "Jane Doe"
24
)
25
print(result)

1
import os
2
3
def generate_subtitles(video_file):
4
    # Extract audio from video (requires ffmpeg)
5
    audio_file = video_file.replace(".mp4", ".mp3")
6
    os.system(f'ffmpeg -i "{video_file}" -q:a 0 -map a "{audio_file}" -y')
7
    
8
    # Transcribe
9
    with open(audio_file, "rb") as f:
10
        transcription = client.audio.transcriptions.create(
11
            model="whisper-1",
12
            file=f,
13
            response_format="srt"
14
        )
15
    
16
    # Save subtitles
17
    srt_file = video_file.replace(".mp4", ".srt")
18
    with open(srt_file, "w", encoding="utf-8") as f:
19
        f.write(transcription.text)
20
    
21
    # Clean up temporary audio
22
    os.remove(audio_file)
23
    
24
    print(f"✅ Subtitles generated: {srt_file}")
25
26
generate_subtitles("presentation.mp4")

1
from pydub import AudioSegment
2
3
def transcribe_large_file(file_path):
4
    # Load audio
5
    audio = AudioSegment.from_file(file_path)
6
    
7
    # Split into 10-minute chunks
8
    chunk_length_ms = 10 * 60 * 1000
9
    chunks = [audio[i:i + chunk_length_ms] 
10
              for i in range(0, len(audio), chunk_length_ms)]
11
    
12
    full_transcript = ""
13
    
14
    for i, chunk in enumerate(chunks):
15
        # Export chunk
16
        chunk_file = f"temp_chunk_{i}.mp3"
17
        chunk.export(chunk_file, format="mp3")
18
        
19
        # Transcribe
20
        with open(chunk_file, "rb") as f:
21
            transcription = client.audio.transcriptions.create(
22
                model="whisper-1",
23
                file=f
24
            )
25
        
26
        full_transcript += transcription.text + " "
27
        
28
        # Clean up
29
        os.remove(chunk_file)
30
    
31
    return full_transcript.strip()

1
def safe_transcribe(audio_path):
2
    try:
3
        with open(audio_path, "rb") as audio_file:
4
            transcription = client.audio.transcriptions.create(
5
                model="whisper-1",
6
                file=audio_file,
7
                response_format="text"
8
            )
9
        return transcription.text
10
    
11
    except FileNotFoundError:
12
        print(f"❌ File not found: {audio_path}")
13
        return None
14
    
15
    except Exception as e:
16
        print(f"❌ Transcription error: {e}")
17
        return None