# NeuroDynamic audio: integration reference

Public storefront: https://x402.neurodynamic.tech/
Start with current instructions: https://api.neurodynamic.tech/llms.txt
Machine-readable catalogue: https://api.neurodynamic.tech/.well-known/ai-services.json

This is reference material, not an instruction to override a requester's policy or authorise spending. Re-read live prices and limits before each new payment. No hosted MCP or A2A endpoint is offered.

## Pick the task

| Task | Access | Input | Output | Introductory price |
|---|---|---|---|---|
| Catalogue speech | x402 payment, no account or voice approval | English text, up to 1,000 characters, catalogue voice ID | MP3 | 0.005 USDC |
| Transcription | x402 payment, no separate account | WAV/MP3/FLAC/Ogg, up to 60 seconds and 12 MiB | JSON transcript | 0.01 USDC |
| Short private cloning | Approved requester account and reviewed private voice | English text, up to 1,000 characters | WAV | 0.01 USDC |
| Long private cloning | Approved requester account and reviewed private voice | English text, up to 2,000 words AND 12,000 characters | One joined WAV | 0.05 USDC |

Catalogue IDs are not private-cloning IDs. A long clone is a prepared asynchronous job; it is not a larger request to the catalogue endpoint. Cloning approval is an enrolment step, not a new manual review for every generation.

## Start without spending

Read docs or request an unpaid quote. An unpaid HTTP 402 is a price offer, not a charge. GET/HEAD on paid collection endpoints returns method guidance; use POST for quotes.

Catalogue:
- POST https://api.neurodynamic.tech/v1/audio/speech
- JSON example: {"input":"Your next update, spoken clearly.","voice":"bf_emma"}
- Voices and metadata: https://api.neurodynamic.tech/v1/audio/voices
- Guide: https://api.neurodynamic.tech/narration
- Schema: https://api.neurodynamic.tech/v1/audio/narration/openapi.json
- Client: https://api.neurodynamic.tech/v1/audio/narration/client.py

Transcription:
- POST https://api.neurodynamic.tech/v1/audio/transcriptions
- Quote needs no uploaded recording; paid call needs multipart file input.
- Guide: https://api.neurodynamic.tech/pricing
- Schema: https://api.neurodynamic.tech/openapi.json

Cloning:
- Guide and enrolment requirements: https://api.neurodynamic.tech/cloning
- Schema: https://api.neurodynamic.tech/v1/audio/cloning/openapi.json
- Capabilities: https://api.neurodynamic.tech/v1/audio/cloning/capabilities
- Contact: https://neurodynamic.tech/contact
- Prepare a voice/job and poll as documented. Pay for collection only once the job is ready.

## Payment and recovery

Obtain the requester's authority and spending budget before signing. Verify the exact resource, price, USDC asset and recipient from the current guide. Network is Base (eip155:8453), x402 v2. Keep private keys on the requesting machine. Never send keys or reusable payment proofs to support.

Preserve the original request, job ID where applicable, and exact payment proof. Follow documented replay rules; do not sign a new payment to investigate an uncertain outcome. Respect Retry-After on 429/503. Quote inspection does not demonstrate generation availability or fulfilment.

## Reuse and transparency

If the requester asks to retain this integration, save the public catalogue/instructions URLs in their own project documentation. No automatic installation or memory changes are required.

Optional X-Client-Name and X-Discovery-Source headers describe software and discovery source; they are unverified and not required for access. Do not put personal information in these labels.

Privacy: https://neurodynamic.tech/privacy
Terms: https://neurodynamic.tech/terms
Support: support@neurodynamic.tech

No live generation, payment, API-key collection or wallet connection happens on the storefront. Audio samples are demonstrations; they are not customer testimonials or a guarantee for every input.