Task-Specific APIs
Task-Specific APIs
Use Soynade's task-specific APIs when you want capability-shaped request and response bodies, multipart local file upload, async jobs, batch processing, or product-specific controls.
These routes do not take a model parameter. Soynade selects the appropriate
production model for the task.
Set your API key first:
Code
Text Translation
Translate between Wolof (wo), English (en), and French (fr) in either
direction. Set source_language and target_language explicitly; the defaults
are Wolof to French. Both /v1/translations and /v1/translations/jobs accept
Wolof as the target. For example, use source_language: "en" and
target_language: "wo" to translate English into Wolof.
The example below translates Wolof text to French.
Speech Recognition
Transcribe a local audio file. Use multipart upload for local wav, mp3, or
flac files.
Speech Translation
Translate speech audio into text in another language.
Text To Speech
Generate speech from text. Save synchronous audio responses with --output so
binary data is not printed to your terminal.
Set output_format to wav (the default) or mp3. WAV responses use
audio/wav; MP3 responses use audio/mpeg. For MP3, set "output_format": "mp3"
in the JSON examples below and save to speech.mp3. With multipart requests,
use -F "output_format=mp3". The reference recording's format does not determine
the output format.
Text To Speech With Reference Audio
Use reference audio only when you want to guide the voice style from a sample
you have permission to use. Reference-audio requests must include
voice_rights_attested=true and a voice_rights_basis of self or
authorized.
For longer workloads, use the async job endpoints. See Async Jobs for separate submit, status, and result-download examples.