zen3-audio

Best Quality Transcription

A 1.5B parameter speech-to-text model delivering the highest transcription accuracy in the Zen family. Supports 100+ languages with strong performance across accents, noisy environments, and domain-specific terminology.

Specifications

Capabilities

• Highest accuracy transcription across 100+ languages
• Speaker diarization (who spoke when)
• Word and segment-level timestamps
• Punctuation and capitalization restoration
• Noise-robust transcription in challenging environments
• Domain-specific vocabulary support (medical, legal, technical)

API Usage

curl https://api.hanzo.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -F model=zen3-audio \
  -F file=@meeting.mp3 \
  -F language=en \
  -F response_format=verbose_json

from hanzoai import Hanzo

client = Hanzo(api_key="hk-your-api-key")

with open("meeting.mp3", "rb") as audio_file:
    response = client.audio.transcriptions.create(
        model="zen3-audio",
        file=audio_file,
        language="en",
        response_format="verbose_json",
        timestamp_granularities=["word", "segment"],
    )

for segment in response.segments:
    print(f"[{segment.start:.1f}s - {segment.end:.1f}s] {segment.text}")

Try It

Open in Hanzo Chat

Resources

• Audio API -- Endpoint documentation
• Technical Report

See Also

• zen3-audio-fast -- Fastest transcription for high throughput
• zen3-asr -- Real-time streaming speech recognition
• zen3-tts -- High-quality text-to-speech