SDK

Clients

Client	Alias	Use it for
`TeenageWorks`	`TW`	Synchronous applications
`AsyncTeenageWorks`	`AsyncTW`	Async applications

Create one client per process or worker and reuse it. This keeps one connection pool available across requests.

from teenageworks import AsyncTeenageWorks

async with AsyncTeenageWorks() as client:
    speech = await client.audio.speech.create(
        model="teen-v1",
        voice="default",
        input="Generate asynchronously.",
        response_format="mp3",
    )
    await speech.write_to_file("speech.mp3")

Resources

Resource	Purpose
`audio.speech`	Create or stream speech
`voices.list()`	List public voices
`models.list()`	List available models
`capabilities.retrieve()`	Inspect formats and supported features
`account.usage.retrieve()`	Read project usage and balance

Response modes

Normal methods return parsed results.
with_raw_response also exposes status, headers, and request ID.
with_streaming_response keeps audio incremental and must be closed or used as a context manager.

Use close() when you do not use a context manager.

Errors and retries

All SDK errors inherit from TeenageWorksError, also available as TWError.

Condition	Error
Invalid key	`AuthenticationError`
Insufficient credit	`BillingError`
Invalid request	`UnprocessableEntityError`
Rate or concurrency limit	`RateLimitError`
Connection failure	`APIConnectionError`
Timeout	`APITimeoutError`

Connection failures, rate limits, and selected server failures may retry. Authentication, billing, validation, and unsupported-feature errors do not. A stream is never restarted automatically after audio bytes have been delivered.