List available models

gethttps://api.dev.developers.gaudiolab.io/v1/models

Returns every separation model your project can use, identified by a stable, human-readable alias. Pass an alias as the model field when creating a job target. Public aliases include a family prefix and version suffix, so similarly named tracks from different model families remain distinct.

The list is scoped to the features enabled for your workspace: a model whose category is not enabled is omitted here and refused by POST /jobs with 403 FEATURE_NOT_ENABLED. A model excluded by the workspace model allowlist is also omitted and refused with 403 MODEL_NOT_ENABLED. For DME models, tiers contains only the tiers enabled for the workspace; a model with no enabled tiers is omitted. Direct use of an excluded tier is refused with 403 TIER_NOT_ENABLED. Each model includes its effective creditMultiplier; workspace overrides replace global defaults. Pricing lookup failures return 503 MODEL_POLICY_UNAVAILABLE. Accepted jobs retain the multiplier captured at admission.

get/models

curl -X GET "https://api.dev.developers.gaudiolab.io/v1/models" \
  -H "x-api-key: $API_KEY"

Response200

{
  "models": [
    {
      "alias": "dme_dialogue_v1",
      "category": "dme_separation",
      "description": "Extracts all voices, including dialogue and vocals, while removing background music and effects.",
      "outputs": [
        "dialogue"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "dme_music_v1",
      "category": "dme_separation",
      "description": "Extracts music, including vocals, while removing dialogue and effects.",
      "outputs": [
        "music"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "dme_music_effects_remove_vocals_v2",
      "category": "dme_separation",
      "description": "Preserves music and effects only, removing both dialogue and vocals, with cleaner separation and higher fidelity than v1.",
      "outputs": [
        "music_effects_remove_vocals"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "stem_bass_v1",
      "category": "stem_separation",
      "description": "Extracts the bass stem from music.",
      "outputs": [
        "bass"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "stem_vocal_v1",
      "category": "stem_separation",
      "description": "Extracts the vocal stem from the multi-stem separation family.",
      "outputs": [
        "vocal"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "karaoke_vocal_v1",
      "category": "karaoke_separation",
      "description": "Extracts the vocal track from the karaoke-style vocal/accompaniment split.",
      "outputs": [
        "vocal"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "denoise_speech_v1",
      "category": "denoise",
      "description": "Isolates clean speech from background noise.",
      "outputs": [
        "speech"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1"
    },
    {
      "alias": "dme_dialogue_remove_vocals_v1",
      "category": "dme_separation",
      "description": "Dialogue-only extraction that removes vocals, music, and effects.",
      "outputs": [
        "dialogue_remove_vocals"
      ],
      "tiers": [
        "premium",
        "standard",
        "live"
      ],
      "formats": [
        "wav"
      ],
      "creditMultiplier": 1,
      "billingVersion": "rounded_minutes_v1",
      "input_channel_layouts": [
        "mono",
        "stereo",
        "5.1"
      ],
      "tier_overrides": {
        "live": {
          "input_channel_layouts": [
            "mono",
            "stereo"
          ]
        }
      }
    }
  ]
}

Headers

x-api-keyRequired

Your project API key, sent on every request. Create and manage keys in the dashboard under Settings → API Keys.

Response

200The workspace-authorized subset of the catalog in spec/models.yaml, with effective pricing from the persisted model policy.

modelsModel[]Optional

creditMultipliernumberOptional

Effective credit multiplier. Each successful target costs ceil(ceil(sourceDurationMs / 60000) * creditMultiplier) credits. Failed targets are not charged.

billingVersionenumOptional

Version of the credit rounding rule.

Allowed values: rounded_minutes_v1

aliasstringRequired

Stable identifier to pass as model when creating a target. Public aliases include a family prefix and version suffix, such as dme_dialogue_v1, stem_vocal_v1, or karaoke_vocal_v1.

categoryenumRequired

High-level family the model belongs to.

Allowed values: dme_separationstem_separationkaraoke_separationdenoisetext_sync

descriptionstringOptional

A short, human-readable summary of what the model extracts.

outputsstring[]Required

The stem keys this model produces. These appear as the first level of keys in a completed target's output object.

tiersenum[]Optional

DME processing tiers available to the authenticated workspace. Present for DME models only; use one of these values as the target tier. Workspace policy may narrow the model's supported tiers.

tier_overridesmap<string, object>Optional

Capabilities that differ for one tier of this model, keyed by tier id. A tier runs its own model, so it can accept less than the model-level fields advertise — a listed field REPLACES the model-level value for that tier only. Present only where a tier actually differs; absent tiers use the model-level values. Today the only overridden field is input_channel_layouts.

formatsenum[]Required

Audio formats this model can render.

input_channel_layoutsenum[]Optional

Input channel layouts this model accepts. Omitted for models that accept only mono and stereo; DME models also list 5.1. A 5.1 input is accepted only when every target in the job accepts 5.1, and the separated tracks are returned as 5.1.

creditsPerMinutenumberOptional

Credits consumed per minute of input audio.

503Model access or pricing could not be loaded (MODEL_POLICY_UNAVAILABLE). Retry later; no fallback pricing is used.

codestringRequired

A stable, machine-readable error code you can branch on.

messagestringRequired

A human-readable explanation of what went wrong.

detailsobjectOptional

Machine-readable specifics for the errors that carry them — for example FEATURE_NOT_ENABLED reports the model that was refused and the feature it belongs to. Absent for errors that add nothing beyond code and message.