Namespace Mabs.AudioCpp.Models
Classes
- AlignmentRequest
Request of
POST /v1/audio/alignments: forced alignment of a known transcript against uploaded audio.
- AlignmentResult
Result of
POST /v1/audio/alignments.
- ArtifactKind
The
kindof an artifact. Wire values accepted by the server.
- AudioArtifact
An opaque model artifact (speaker embedding, acoustic tokens, …) returned by tasks routes.
- AudioArtifactInput
An artifact sent in a task request (
request.artifacts[]).
- AudioCppHealth
Server readiness from
GET /health.
- AudioCppModel
A model entry from
GET /v1/models.
- AudioCppModelParam
A request option declared by a model spec, from
GET /v1/models?include_params=true.
- AudioCppSegment
A speech segment reported by a model (timestamps are sample offsets).
- AudioCppSpeakerTurn
A speaker turn from a diarization model.
- AudioCppTiming
Inference timing. Different endpoints populate different members: non-streaming runs report WallMs, AudioDurationMs and Rtf; streaming runs report TtftMs (when the server sent a
ttft_mskey — check TtftPresent because a live-speech timing can carry an explicit JSONnull); live speech adds the remaining members.
- AudioCppWord
A word with timestamps, from alignment or word-level transcription output.
- AudioUpload
One uploaded file (multipart part).
- BatchTranscriptionRequest
Request of
POST /v1/batches/transcriptions: multiple uploaded WAV files through one native offline batch. The model must implement native batching.
- LiveSpeechRequest
Parameters of
POST /v1/audio/speech/live: chunked PCM body in, SSE or raw audio out. All request fields travel as query parameters because the body carries audio.
- LiveTranscriptionRequest
Query parameters of
POST /v1/audio/transcriptions/live. The body is raw interleaved PCM sent chunked; there is no multipart form and no JSON.
- ModelLoadRequest
Body of
POST /v1/models/load(builds with native model management).
- ModelLoadResult
Result of
POST /v1/models/load.
- NamedAudioOutput
One named audio output of a result or stream event.
- RecipeEntry
A recipe entry from
GET /v1/recipes.
- RecipeRole
One required role of a recipe.
- RecipeRunRequest
Body of
POST /v1/recipes/run.
- SpeechJsonResult
Parsed
response_format=jsonspeech response.
- SpeechRequest
Body of
POST /v1/audio/speech(non-streaming) and the base of stream requests.
- SpeechResult
Non-streaming speech result: audio bytes, content type, and timing headers.
- SpeechStreamRequest
Body of a streaming speech request (
stream_formatalways present).
- TaskBatchRequest
Body of
POST /v1/tasks/batch.
- TaskRequestBody
The
requestobject accepted byPOST /v1/tasks/run,/v1/tasks/stream(as each entry of/v1/tasks/batch). It mirrors the audio.cpp CLI request-sequence format: exactly one of AudioPath / AudioBase64 for audio input; family-specific parameters go in Options (values serialize the way the server's option map does: boolstrue/false, integers without decimals, doubles invariant).
- TaskResult
A result of
POST /v1/tasks/run,/v1/tasks/streamor a recipe run.
- TaskRunRequest
Body of
POST /v1/tasks/run.
- TaskStreamEventBody
One event of a task stream (both the JSON
events[]rows and the SSEtask.stream.eventpayload).
- TaskStreamRequest
Body of
POST /v1/tasks/stream.
- TranscriptionDetailsRequest
Request of
POST /v1/audio/transcriptions/details. Streaming is rejected by the server on this route.
- TranscriptionDetailsResult
Rich transcription response with the detail arrays the model produced.
- TranscriptionInputBase
Input for the transcription routes: either a server-local WAV path (AudioPath, sent as JSON) or uploaded audio bytes (AudioFile + AudioFileName, sent as multipart). Exactly one form must be set.
- TranscriptionRequest
Request of
POST /v1/audio/transcriptions.
- TranscriptionResult
Non-streaming transcription response (
{text, timing}).
- UnloadModelsResult
Result of
POST /v1/tasks/unload_modelsand/v1/tasks/unload_all_models.
- VoiceRefObject
A
voice_refobject:{"type":"path","path":…}or{"type":"base64","data":…}(decoded payload limited to 5 MiB by the server). Plain server-local path strings are also accepted by the wire; see VoiceRef.
Structs
- BatchTranscriptionEvent
A batch transcription stream event.
- ResponseFormat
A
response_formatvalue. The wire type is a string; use the static instances or construct from any string the server accepts (an unknown value is rejected by the server).
- SpeechStreamEvent
A speech stream event.
- StreamFormat
A
stream_formatvalue:sseevents or raw chunked audio.
- TaskBatchEvent
A task batch stream event.
- TaskStreamEvent
A task stream event: an incremental Event or the terminal Result.
- TranscriptEvent
A transcription stream event.
Enums
- BatchTranscriptionEventKind
Kind of a batch transcription stream event.
- SpeechStreamEventKind
Kind of a SpeechStreamEvent.
- TaskBatchEventKind
Kind of a task batch stream event.
- TaskStreamEventKind
Kind of a task stream event surfaced by the client.
- TranscriptEventKind
Kind of a transcription stream event.