Audio Translation

1
from openai import OpenAI
2

3
client = OpenAI(
4
    base_url="https://api.neura-ai.app/v1"
5
)
6

7
with open("spanish_audio.mp3", "rb") as audio_file:
8
    translation = client.audio.translations.create(
9
        model="whisper-1",
10
        file=audio_file
11
    )
12

13
print(translation.text)

1
with open("french_audio.mp3", "rb") as audio_file:
2
    translation = client.audio.translations.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        response_format="text"
6
    )
7

8
print(translation.text)

1
with open("german_audio.mp3", "rb") as audio_file:
2
    translation = client.audio.translations.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        response_format="json"
6
    )
7

8
print(translation.text)

1
with open("italian_audio.mp3", "rb") as audio_file:
2
    translation = client.audio.translations.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        response_format="verbose_json"
6
    )
7

8
print(f"Duration: {translation.duration} seconds")
9

10
for segment in translation.segments:
11
    print(f"[{segment.start:.2f}s - {segment.end:.2f}s]: {segment.text}")

1
# SRT format
2
with open("japanese_video.mp3", "rb") as audio_file:
3
    translation = client.audio.translations.create(
4
        model="whisper-1",
5
        file=audio_file,
6
        response_format="srt"
7
    )
8

9
with open("english_subtitles.srt", "w") as f:
10
    f.write(translation.text)
11

12
# VTT format
13
with open("korean_video.mp3", "rb") as audio_file:
14
    translation = client.audio.translations.create(
15
        model="whisper-1",
16
        file=audio_file,
17
        response_format="vtt"
18
    )
19

20
with open("english_subtitles.vtt", "w") as f:
21
    f.write(translation.text)

1
with open("chinese_business.mp3", "rb") as audio_file:
2
    translation = client.audio.translations.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        prompt="Business meeting discussing quarterly sales and marketing strategy"
6
    )
7

8
print(translation.text)

1
with open("russian_lecture.mp3", "rb") as audio_file:
2
    translation = client.audio.translations.create(
3
        model="whisper-1",
4
        file=audio_file,
5
        temperature=0.0  # Most consistent/deterministic
6
    )

1
def translate_news_clip(audio_file, topic):
2
    with open(audio_file, "rb") as f:
3
        translation = client.audio.translations.create(
4
            model="whisper-1",
5
            file=f,
6
            response_format="verbose_json",
7
            prompt=f"News report about {topic}"
8
        )
9
    
10
    # Save translated transcript
11
    output_file = audio_file.replace(".mp3", "_english.txt")
12
    with open(output_file, "w") as f:
13
        f.write(f"Topic: {topic}\n")
14
        f.write(f"Duration: {translation.duration:.2f}s\n\n")
15
        f.write(translation.text)
16
    
17
    return translation.text
18

19
translate_news_clip("french_news.mp3", "European Union policy")

1
def create_english_subtitles(video_audio, original_language):
2
    with open(video_audio, "rb") as f:
3
        translation = client.audio.translations.create(
4
            model="whisper-1",
5
            file=f,
6
            response_format="srt"
7
        )
8
    
9
    subtitle_file = video_audio.replace(".mp3", "_EN.srt")
10
    with open(subtitle_file, "w", encoding="utf-8") as f:
11
        f.write(translation.text)
12
    
13
    print(f"✅ English subtitles created: {subtitle_file}")
14
    return subtitle_file
15

16
create_english_subtitles("spanish_tutorial.mp3", "Spanish")

1
def translate_support_call(call_recording, customer_language):
2
    with open(call_recording, "rb") as f:
3
        translation = client.audio.translations.create(
4
            model="whisper-1",
5
            file=f,
6
            response_format="verbose_json",
7
            prompt="Customer support call discussing technical issues"
8
        )
9
    
10
    # Create formatted transcript
11
    transcript = f"Support Call Translation\n"
12
    transcript += f"Original Language: {customer_language}\n"
13
    transcript += f"Duration: {translation.duration:.2f}s\n"
14
    transcript += "=" * 50 + "\n\n"
15
    
16
    for segment in translation.segments:
17
        timestamp = f"[{int(segment.start//60):02d}:{int(segment.start%60):02d}]"
18
        transcript += f"{timestamp} {segment.text}\n"
19
    
20
    return transcript
21

22
result = translate_support_call("german_support.wav", "German")
23
print(result)

1
def translate_lecture(lecture_file, subject):
2
    with open(lecture_file, "rb") as f:
3
        translation = client.audio.translations.create(
4
            model="whisper-1",
5
            file=f,
6
            response_format="text",
7
            prompt=f"University lecture on {subject}",
8
            temperature=0.1  # More consistent for educational content
9
        )
10
    
11
    # Save as markdown for easy reading
12
    output_file = lecture_file.replace(".mp3", "_english.md")
13
    with open(output_file, "w") as f:
14
        f.write(f"# Lecture: {subject}\n\n")
15
        f.write(translation.text)
16
    
17
    return translation.text
18

19
translate_lecture("physics_lecture_french.mp3", "Quantum Mechanics")

1
def translate_podcast_episode(episode_file, show_name, episode_num):
2
    with open(episode_file, "rb") as f:
3
        translation = client.audio.translations.create(
4
            model="whisper-1",
5
            file=f,
6
            response_format="verbose_json",
7
            prompt=f"Podcast: {show_name}, Episode {episode_num}"
8
        )
9
    
10
    # Create formatted English transcript
11
    transcript = f"# {show_name} - Episode {episode_num}\n"
12
    transcript += f"## English Translation\n\n"
13
    transcript += translation.text
14
    
15
    # Save
16
    output_file = f"{show_name}_ep{episode_num}_EN.md"
17
    with open(output_file, "w") as f:
18
        f.write(transcript)
19
    
20
    print(f"✅ Translated podcast saved to {output_file}")
21
    return transcript
22

23
translate_podcast_episode(
24
    "italian_podcast.mp3",
25
    "Tech Talk Italy",
26
    "042"
27
)

1
# One step - Direct translation
2
with open("french.mp3", "rb") as f:
3
    result = client.audio.translations.create(model="whisper-1", file=f)

1
# Two steps - Transcribe then translate
2
# Step 1: Transcribe in original language
3
with open("french.mp3", "rb") as f:
4
    transcription = client.audio.transcriptions.create(
5
        model="whisper-1", 
6
        file=f,
7
        language="fr"
8
    )
9

10
# Step 2: Translate text (requires text translation API)
11
# (This would require additional API call)

1
from pydub import AudioSegment
2
import os
3

4
def translate_large_file(file_path):
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_translation = ""
13
    
14
    for i, chunk in enumerate(chunks):
15
        chunk_file = f"temp_chunk_{i}.mp3"
16
        chunk.export(chunk_file, format="mp3")
17
        
18
        with open(chunk_file, "rb") as f:
19
            translation = client.audio.translations.create(
20
                model="whisper-1",
21
                file=f
22
            )
23
        
24
        full_translation += translation.text + " "
25
        os.remove(chunk_file)
26
    
27
    return full_translation.strip()

1
def safe_translate(audio_path, context=""):
2
    try:
3
        with open(audio_path, "rb") as f:
4
            translation = client.audio.translations.create(
5
                model="whisper-1",
6
                file=f,
7
                prompt=context
8
            )
9
        return translation.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"❌ Translation error: {e}")
17
        return None
18

19
result = safe_translate("german_audio.mp3", "Technical presentation")
20
if result:
21
    print(result)