API-Referenz / AI utilities
AI utilities
Die Endpoints der Gruppe AI utilities der Voiceland AI API, mit Parametern, Schemata und Beispielen mit curl.
Zuletzt aktualisiert:
Die Beschreibungen der Endpoints und der Felder erscheinen auf Englisch, genau so, wie die API sie ausliefert. Das ist Absicht: Sie lesen hier, was Sie auch in den Antworten sehen.
POST /v1/chat#
Ask a question over a transcript. Answers a question about a transcript. freeform returns an answer; structured also returns matched + a summary.
Anfragetext
application/json Schema: ChatRequest
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
history |
array of ChatMessage |
Optional prior turns for multi-turn context. | |
max_tokens |
integer | Optional cap on the response length. | |
mode |
string | Ja | freeform (default) returns an answer; structured also returns matched + a summary. |
model |
string | Optional model override from the account catalog. | |
question |
string | Ja | The question to answer about the transcript. |
transcript |
string | Ja | The conversation text to reason over. |
Antworten
| Code | Beschreibung |
|---|---|
200 |
Success. |
401 |
Missing or invalid API key. |
4XX |
Request error (validation, not-found, etc.). |
5XX |
Server or upstream error. |
Beispielanfrage
curl -X POST "https://api.voiceland.ai/v1/chat" \
-H "Authorization: Bearer VL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"mode": "structured",
"question": "What order number?",
"transcript": "caller: cancel order 4471\nagent: done."
}'Eine erfolgreiche Antwort liefert ein StructuredResponse.
Beispielantwort
{
"answer": "4471",
"matched": true,
"summary": "Caller cancelled order 4471."
}POST /v1/transcribe#
Transcribe audio. Transcribes an uploaded audio file. Send multipart/form-data with a file part; optional language, model, and speaker_diarization fields.
Anfragetext
multipart/form-data
Antworten
| Code | Beschreibung |
|---|---|
200 |
Success. |
401 |
Missing or invalid API key. |
4XX |
Request error (validation, not-found, etc.). |
5XX |
Server or upstream error. |
Beispielanfrage
curl -X POST "https://api.voiceland.ai/v1/transcribe" \
-H "Authorization: Bearer VL_API_KEY" \
-F "file=@audio.wav" \
-F "language=el" \
-F "speaker_diarization=true"Beispielantwort
{
"audio_duration_seconds": 12.3,
"language": "el",
"text": "Καλησπέρα…",
"transcription_id": "…"
}Schemata#
ChatMessage#
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
content |
string | Ja | |
role |
string | Ja |
ChatRequest#
Ask a question over a transcript.
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
history |
array of ChatMessage |
Optional prior turns for multi-turn context. | |
max_tokens |
integer | Optional cap on the response length. | |
mode |
string | Ja | freeform (default) returns an answer; structured also returns matched + a summary. |
model |
string | Optional model override from the account catalog. | |
question |
string | Ja | The question to answer about the transcript. |
transcript |
string | Ja | The conversation text to reason over. |
ChatUsage#
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
completion_tokens |
integer | Ja | |
prompt_tokens |
integer | Ja | |
total_tokens |
integer | Ja |
Error#
Error envelope returned for non-2xx responses.
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
error |
object | Ja |
StructuredResponse#
| Feld | Typ | Erforderlich | Beschreibung |
|---|---|---|---|
answer |
string | Ja | The extracted answer. |
matched |
boolean | Ja | Whether the transcript contained an answer. |
model |
string | Ja | |
summary |
string | Ja | A short summary of the transcript. |
usage |
ChatUsage |
Ja |