Configuration Reference
Configuration schema for Persyk application settings
Type: object
⚠️ Additional properties are not allowed.
| Property | Type | Required | Possible values | Deprecated | Default | Description | Examples |
|---|---|---|---|---|---|---|---|
| providers | object | ✅ | object | {} | API provider configurations | ||
| transcription | object | object | Transcription settings | ||||
| transcription.enabled | boolean | ✅ | boolean | true | Whether transcription is enabled | ||
| transcription.provider | string | ✅ | string | Name of the provider to use for transcription | |||
| transcription.model | string | ✅ | string | "gpt-4o-mini-transcribe" | Realtime transcription model name | ||
| transcription.language | string | string | Language code for transcription (e.g., ‘en’, ‘es’) | ||||
| transcription.keywords | string[] | array of strings | Custom vocabulary words or phrases; OpenAI supports gpt-transcribe and gpt-live-transcribe | ["Persyk", "SvelteKit"] | |||
| transcription.turnDetection | object | ✅ | InlineObject1 or InlineObject2 | {"type": "server_vad", "threshold": 0.5, "prefix_padding_ms": 300, "silence_duration_ms": 500} | Turn detection configuration for the Realtime API | ||
| transcription.noiseReduction | object | ✅ | object | {"type": "near_field"} | Noise reduction configuration | ||
| transcription.noiseReduction.type | string | ✅ | near_field far_field | Noise reduction type based on microphone distance | |||
| transcription.copyToClipboard | boolean | ✅ | boolean | true | Copy transcription to clipboard after completion | ||
| transcription.notifications | boolean | ✅ | boolean | true | Show notifications for transcription events |
Definitions
InlineObject1
Server-side VAD turn detection
Type: object
⚠️ Additional properties are not allowed.
| Property | Type | Required | Possible values | Deprecated | Default | Description | Examples |
|---|---|---|---|---|---|---|---|
| type | const | ✅ | server_vad | Server-side Voice Activity Detection | |||
| threshold | number | ✅ | 0 <= x <= 1 | 0.5 | VAD activation threshold (0-1) | ||
| prefix_padding_ms | number | ✅ | number | 300 | Milliseconds of audio to include before speech starts | ||
| silence_duration_ms | number | ✅ | number | 500 | Milliseconds of silence before speech is considered ended |
InlineObject2
Semantic VAD turn detection
Type: object
⚠️ Additional properties are not allowed.
| Property | Type | Required | Possible values | Deprecated | Default | Description | Examples |
|---|---|---|---|---|---|---|---|
| type | const | ✅ | semantic_vad | Semantic Voice Activity Detection | |||
| eagerness | string | ✅ | low medium high | "medium" | How eagerly to detect end of turn |
Markdown generated with jsonschema-markdown.