Voice AI and Transcription

import openai, os

client = openai.OpenAI(api_key=os.environ["OPENAI_API_KEY"])

def transcribe_audio(file_path: str, language: str = None) -> dict:
    with open(file_path, "rb") as f:
        transcript = client.audio.transcriptions.create(
            model="whisper-1",
            file=f,
            language=language,       # ISO 639-1 code, e.g. "en"
            response_format="verbose_json",
            timestamp_granularities=["segment"]
        )
    return {
        "text": transcript.text,
        "duration": transcript.duration,
        "segments": [{"start": s.start, "end": s.end, "text": s.text}
                      for s in (transcript.segments or [])]
    }

# result = transcribe_audio("meeting.mp3")
# print(result["text"][:200])

def summarize_meeting(transcript_text: str) -> dict:
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "system", "content": """Analyze this meeting transcript and provide:
{
  "summary": "3-5 sentence overview",
  "key_decisions": ["decision 1", ...],
  "action_items": [{"owner": "name", "task": "description", "deadline": "if mentioned"}],
  "topics_discussed": ["topic 1", ...],
  "follow_ups_needed": ["item 1", ...]
}
Reply with JSON only."""}, {"role": "user", "content": transcript_text[:12000]}],
        response_format={"type": "json_object"},
        temperature=0.2
    )
    return json.loads(response.choices[0].message.content)

# Full pipeline
def process_meeting_recording(audio_path: str) -> dict:
    print("Transcribing...")
    transcript = transcribe_audio(audio_path)
    print(f"Transcription: {len(transcript['text'])} chars, {transcript['duration']:.0f}s")

    print("Summarizing...")
    summary = summarize_meeting(transcript["text"])
    return {"transcript": transcript, "summary": summary}

# meeting = process_meeting_recording("standup.mp3")

def text_to_speech(text: str, output_path: str = "output.mp3",
                     voice: str = "alloy", speed: float = 1.0):
    """Voices: alloy, echo, fable, onyx, nova, shimmer"""
    response = client.audio.speech.create(
        model="tts-1",       # or tts-1-hd for higher quality
        voice=voice,
        input=text,
        speed=speed           # 0.25 to 4.0
    )
    response.stream_to_file(output_path)
    return output_path

# Generate audio from article text
# text_to_speech("Welcome to our weekly product update...", "update.mp3", voice="nova")

# Batch: convert articles to podcast
def articles_to_podcast(articles: list[dict], output_dir: str = "podcast"):
    os.makedirs(output_dir, exist_ok=True)
    for i, article in enumerate(articles):
        # Create a podcast script
        script_resp = client.chat.completions.create(
            model="gpt-4o-mini",
            messages=[{"role": "system", "content": "Convert this article into a natural podcast script. Add transitions, keep it conversational. 2-3 minutes of speech."}, {"role": "user", "content": article["content"][:4000]}],
            temperature=0.7
        )
        script = script_resp.choices[0].message.content
        output = os.path.join(output_dir, f"ep{i+1}_{article['slug']}.mp3")
        text_to_speech(script, output, voice="nova")
        print(f"Generated: {output}")