Diffio API

Build audio restoration into your product.

Add restoration to your app with one API key and a simple SDK flow.

SDK quickstart

Use the same short flow as the SDK docs.

The helper below creates the project, runs the generation, waits for completion, and downloads audio bytes.

Example restoration

Listen to a restored Amelia Earhart broadcast to hear the API results.

Historical tape

Amelia Earhart

A 1935 broadcast restored.

Loading waveform...
Loading audio enhancer...

Available models

Pick the balance of speed and fidelity that fits your product.

Diffio 2.0 Flash

Fast

Fastest turnaround for previews and quick cleanups.

Model iddiffio-2-flash
  • Low latency option for high volume queues.
  • Great for rapid iterations and early reviews.
  • Balanced cleanup for shorter clips.

Diffio 2.0

Balanced

Reliable baseline model for consistent cleanup.

Model iddiffio-2
  • Steady results for interviews and podcasts.
  • Good balance of speed and quality.
  • Stable output across large batches.

Diffio 3.4

Best

Best quality reconstruction for demanding speech.

Model iddiffio-3.4
  • Improved clarity in noisy rooms.
  • Sharper consonants and tighter detail.
  • Stronger alignment for transcript review.