Overview
Compare Diffio models and choose the right balance of speed and clarity. Get setup details, API behavior, responses, and workflows.
Diffio offers 2 speech restoration models: Diffio 4.5 Flash and Diffio 4.5 Pro. Choose a model for every generation, and run multiple generations on the same project to compare results.
Use the model id in the SDK model field or call the matching REST endpoint. Each model focuses on speech, with different performance and fidelity targets.
Our flagship. Every window is restored three times and only the cleanest result is kept.
Best for: Critical clarity work, difficult recordings, final masters.
Fast, high quality speech restoration for everyday work.
Best for: Everyday restoration, quick turnaround, multilingual speech.
Start with a fast preview, then move to higher fidelity for final delivery. Keep the same project and create a new generation each time you change models.
diffio-4.5-flash for fast, high quality speech restoration when throughput matters.diffio-4.5-pro for the best quality on noisy, distant, or damaged speech.Diffio 4.5 Flash is available on free and paid plans. Diffio 4.5 Pro requires a paid plan in the web app. Review billing details for plan options.
Diffio 4.5 Flash is the fast option and the default on the free plan. Diffio 4.5 Pro delivers the best quality and is the default on paid plans. Both return restored audio, original and restored comparison playback that starts while restoration runs, MP3 downloads, restored video for video projects, and transcript data.
English, French, German, Spanish, and Portuguese are the target languages for quality validation. Other languages still run, with a notice that restoration quality has not been validated for that language. Transcription failure is reported instead of silently proceeding without text guidance.
Requests that name any other model id are rejected with HTTP 410 and the code model_retired. Use diffio-4.5-flash or diffio-4.5-pro. The Python and Node SDKs, from 0.2.0, reject retired model ids before sending a request. Generations created with retired models keep their original modelKey and can still be listed, polled, and downloaded.
Model ids map to specific generation endpoints. Use these paths when calling the REST API directly.
| Model id | Label | Endpoint |
|---|---|---|
diffio-4.5-flash | Diffio 4.5 Flash | POST /v1/diffio-4.5-flash-generation |
diffio-4.5-pro | Diffio 4.5 Pro | POST /v1/diffio-4.5-pro-generation |