Google Lyria2 | Text to Audio

Lyria 2 Description

Google's Lyria 2 is an advanced AI model engineered for high-fidelity music generation and produces professional-grade, instrumental audio from text prompts.

‍Technical Specification

Performance Benchmarks

Lyria 2 is optimized for professional-grade, text-to-music audio generation.

• Audio Output: 30 seconds of 48kHz stereo WAV audio per generation.
• Generation Speed: Audio clips are typically generated within 10 to 20 seconds.

• API Pricing: 0.105 $/generation

Performance Metrics

Despite the absence of formal benchmarks, user feedback subjectively praise Lyria 2 for its superior quality, with some claiming it surpasses competitors like Suno and Udio, and highlight its granular creative controls as a key feature for professional musicians.

Key Capabilities

Lyria 2 delivers precise and controllable audio outputs for creative and professional applications.

• Text-to-Music Generation: Creates instrumental music tracks from descriptive text prompts specifying genre, mood, instrumentation, and tempo.
• Advanced Controls: Supports negative prompting to exclude unwanted elements and seed-based generation for reproducible outputs. Advanced settings for tempo (BPM) and key signature are also available.
• Genre Versatility: Covers a wide range of musical styles, including classical, jazz, pop, electronic, orchestral, and various regional styles.

Optimal Use Cases

• Content Creation: Generating background music and scores for digital content like YouTube Shorts videos.
• Musical Prototyping: Assisting musicians in rapidly creating demos and exploring musical ideas with specific controls over tempo and key.
• Soundscape Design: Creating ambient audio and sound effects for various projects from descriptive text

Code Samples

Music Generation

Get Generated Audio

Parameters

• model: string
• prompt: string - The prompt to generate audio.
• negative_prompt: string - A description of what to exclude from the generated audio
• seed: number - seed for deterministic generation. If provided, the model will attempt to produce the same audio given the same prompt and other parameters.

Comparison with Other Models

• Vs. Google MusicFX/MusicLM: Lyria 2 is an improvement on Google's previous models, offering more advanced controls like BPM and key signature, lyric-to-song generation.
• ‍Vs. Lyria RealTime: Lyria 2 is designed for generating complete audio clips, whereas Lyria RealTime is a separate model focused on interactive and real-time music creation and performance.

API Integration

Accessible via AI/ML API. Documentation: available here.