What is Gemini's native audio understanding capability and how do you use it?

from google import genai
from google.genai import types
from pathlib import Path

client = genai.Client()

# Method 1: Inline audio (small files < 20MB)
audio_bytes = Path("interview.mp3").read_bytes()
response = client.models.generate_content(
    model="gemini-3.5-flash",
    contents=[
        types.Part(
            inline_data=types.Blob(
                data=audio_bytes,
                mime_type="audio/mp3"
            )
        ),
        types.Part(text="Transcribe this interview and summarise the key points.")
    ]
)
print(response.text)

# Method 2: Upload via Files API (recommended for files > 20MB)
uploaded = client.files.upload(
    file=Path("podcast.mp3"),
    config={"display_name": "Tech Podcast Episode 42"}
)

# Wait for processing:
import time
while uploaded.state == "PROCESSING":
    time.sleep(2)
    uploaded = client.files.get(name=uploaded.name)

# Analyse the audio:
response = client.models.generate_content(
    model="gemini-3.5-flash",
    contents=[
        types.Part(file_data=types.FileData(file_uri=uploaded.uri, mime_type="audio/mp3")),
        types.Part(text="How many different speakers are in this podcast? Identify each speaker's role.")
    ]
)
print(response.text)