Custom Models API
Custom speech recognition models for organizations.
Endpoints
| Method | Endpoint | Description |
|---|---|---|
| POST | /api/v3/custom_models | Create custom model |
| GET | /api/v3/speech-to-text/models | List models with pagination |
| GET | /api/v3/custom_models/{id} | Get model details |
| PUT | /api/v3/custom_models/{id} | Update model metadata |
| DELETE | /api/v3/custom_models/{id} | Delete model |
| GET | /api/v3/custom_models/{id}/data | Get model training data |
| POST | /api/v3/custom_models/{id}/run | Start training |
TypeScript Types
CreateCustomModelPayload
type CreateCustomModelPayload = {
name: string;
language_id: number;
organization_id: number;
};
Model
type Model = {
id: number;
key: string;
name: string;
last_modified: string;
organization_id: number;
language_data: null;
is_trainable: boolean;
type: number;
base_language_id: number;
base_language_key: string;
training_log?: {
language_id: number;
training_status: number;
log_name: null;
};
};
CustomModels
type CustomModels = {
count: number;
total_results: number;
result: Model[];
};
UpdateModelBody
type UpdateModelBody = {
name: string;
language_id: number;
organization_id: number;
};
Training Status
The training_status field in training_log indicates the model's training state:
- Training status is polled every 20 seconds during active training
- Status updates are fetched with background refetching enabled
Query Parameters
List models endpoint supports:
offset- Pagination offsetlimit- Results per pageq- Search querysort- Sort fielddirection- Sort direction (asc/desc)
Notes
- Custom models are organization-specific
- Models require a base language ID for initialization
- Training can be triggered via the
/runendpoint - Model details auto-refresh every 20 seconds when viewing
- All models have a unique key identifier
Next Steps
- Create Model - Create your first custom model
- Upload Training Data - Prepare and upload audio + transcripts
- Train Model - Start training and monitor progress
- Data Models - Complete schema reference
Ready to get started? Create your first custom model and achieve higher accuracy for your domain!