Data model

Recordings

One captured audio clip under a parent note. References an S3 blob via data.blob_id, knows its duration_seconds + sample_rate + mime, and carries a free-form transcriptions JSONB keyed by model name (vosk_small, vosk_large, future engines) so we can re-transcribe with a different engine later without losing the prior result. The active transcription is whichever key sits at data.primary_model; the FE renders that one by default but lets the user pick any past run.

Model name: recording

Endpoints: 5

Max page size: 200

Fields

Per-field validation rules. Values that violate any constraint are rejected with 400 before they reach the database.

Field	Type	Constraints
mime	`string`	max length`64`
state	`enum`	enum`uploading \| ready \| transcribing \| transcribed \| failed`
blob_id	`string`	max length`64`
channels	`number`	-
filename	`string`	max length`200`
language	`string`	max length`8`
client_id	`string`	max length`64`
parent_id	`string`	max length`64`ref →`voice_note`owned
size_bytes	`number`	-
recorded_at	`string`	max length`32`
sample_rate	`number`	-
primary_model	`string`	max length`32`
transcriptions	`dict`	-
requested_model	`string`	max length`32`
appended_to_body	`bool`	-
duration_seconds	`number`	-
transcription_error	`string`	max length`400`

Mutability

Which fields can you send, and when? Anything without a marker is server-managed - sending it isn't an error, it's silently ignored.

Create-only - read from POST body.Patchable - read from PATCH body.Server-managed - ignored on the body.

Field	Type	Create	Patch
mime	string
state	enum
blob_id	string
channels	number
filename	string
language	string
client_id	string
parent_id	string
size_bytes	number
recorded_at	string
sample_rate	number
primary_model	string
transcriptions	dict
requested_model	string
appended_to_body	bool
duration_seconds	number
transcription_error	string
transcript_edit	-

Fields marked create-only but not patchable are immutable after creation. Server-managed fields include id, timestamps, ownership, and status.

Filtering & sorting

Combinable on list endpoints. Repeating a filter key produces an IN clause; prefixing a sort key with - reverses direction. Example: ?status=open&status=blocked&sort=-created_at.

Filter keys

parent_iddata__parent_id

statedata__state

primary_modeldata__primary_model

statusstatus

is_archivedis_archived

owned_byowned_by

Sort keys

created_atcreated_at

recorded_atdata__recorded_at

Default: -data__recorded_at

Endpoints

Each endpoint below lists its HTTP method, path, and the PAT scope it needs. Code samples cover curl, JavaScript, TypeScript, Python, Rust, Java, and WebSocket.

GET/xapi2/data/recordingrecording:list

List objects

Returns a paginated list of objects you can read. Default page size is 20; pass ?limit= to change (capped per type). Use ?after=<id> for keyset pagination on created_at-sorted lists, or ?offset= for offset paging.

curl -H "Authorization: Bearer pat_…" \
     "https://sprachmemo.de/xapi2/data/recording?limit=20"

GET/xapi2/data/recording/{id}recording:read

Read one

Returns the object by id. 404 if it does not exist or you cannot read it (the two cases are intentionally conflated).

curl -H "Authorization: Bearer pat_…" \
     https://sprachmemo.de/xapi2/data/recording/OBJECT_ID

POST/xapi2/data/recordingrecording:create

Create

Creates a new object. Body is a flat JSON dict of field values. Server-side fields (id, timestamps, ownership) are filled automatically; only fields listed below as creatable are read from the body.

curl -H "Authorization: Bearer pat_…" \
     -H "Content-Type: application/json" \
     -X POST https://sprachmemo.de/xapi2/data/recording \
     -d '{"name": "…"}'

PATCH/xapi2/data/recording/{id}recording:update

Update

Partial update. Only fields included in the body are touched; everything else is preserved. Same allow-list as create, minus the fields that are immutable post-create.

curl -H "Authorization: Bearer pat_…" \
     -H "Content-Type: application/json" \
     -X PATCH https://sprachmemo.de/xapi2/data/recording/OBJECT_ID \
     -d '{"name": "…"}'

DELETE/xapi2/data/recording/{id}recording:delete

Delete

Removes the object. It vanishes from every default list immediately and stops being returned by read / list.

curl -H "Authorization: Bearer pat_…" \
     -X DELETE https://sprachmemo.de/xapi2/data/recording/OBJECT_ID

Use in CLI

The same endpoints are also exposed via the Sprachmemo CLI. For scripts, CI, and bulk imports it's usually the faster path.

voicecli recording list --limit 5
voicecli recording get <id>
voicecli recording create --parent-id "Hello"
voicecli recording upsert --unique parent_id --csv items.csv
voicecli recording schema   # fields & limits

Full command reference, profiles, CSV import, auto-retry, NDJSON streaming → /docs/cli