Hands-on evaluation
Two paths to test Shunya Labs hands-on before signing an Enterprise contract: self-serve via Pay-as-you-go credit (open to anyone, no sales call required), or a guided Enterprise evaluation with credentials provisioned by Shunya.
Pick your path
Sign up, get an API key, and start calling the cloud endpoints in minutes. Best when you want to validate accuracy or latency on your own audio without scheduling a call.
- $200 of credit, covers roughly 50 hours of ASR or comparable TTS volume.
- No card required, no expiration on the credit.
- Full access to all production models and the Playground.
Shunya provisions a workspace, evaluation credit, and a dedicated point of contact for your team. Best when you need on-prem options, custom volume, or specific compliance review before going to procurement.
- Pre-provisioned evaluation workspace with API keys.
- Custom credit allowance sized to your evaluation plan.
- Direct line to a Shunya solutions engineer for technical questions.
- Optional on-prem trial deployment on hardware you supply (regulated industries).
Path 1: Self-serve
Sign up at playground.shunyalabs.ai. The $200 credit lands on your account automatically, no card on file, no expiration.
From the dashboard, head to API Keys and create a key for your environment. Copy and store it in a secrets manager, Shunya hashes keys at rest, so you can't recover one once you close the modal.
Two fastest paths:
- Without code, open the Playground and try Speech to Text on a sample audio file.
- With code, see the Quickstart for cURL / Python / Node samples.
50 hours of credit is enough to validate accuracy on a realistic sample. Pick 50-200 clips that match your production distribution (codec, language mix, recording conditions), run them through the model you're considering, and measure WER or qualitative accuracy.
Path 2: Enterprise evaluation
Use the contact form ↗ at shunyalabs.ai/contact. Mention you're evaluating for an enterprise deployment and include:
- Use case (BFSI collections, clinical scribe, contact-centre QA, etc.)
- Approximate monthly volume (audio hours or character count)
- Deployment preference (cloud, on-prem, air-gapped)
- Languages required
- Compliance constraints (HIPAA, RBI data localisation, GDPR, etc.)
- Evaluation timeline
A solutions engineer will walk you through the relevant models, intelligence-layer features, and deployment options. By the end of the call you should have agreed: which models you'll test, what counts as success, the credit budget for the evaluation, and the support channel.
Shunya provisions an evaluation workspace and sends:
- API key(s) scoped to the evaluation workspace
- Dashboard / Playground access for your stakeholders
- Credit allocation for the agreed evaluation period
- A direct contact (email or shared Slack) for technical questions
Test against your defined success criteria. Typical evaluation activities:
- Accuracy benchmark on representative audio.
- Latency tests over your real network conditions.
- Compliance review, request the SOC 2 Type II report, ISO 27001 certificate, BAA, and DPA via your account team.
- Integration spike, wire ASR streaming into a sample call flow, or wire TTS into your IVR.
- Cost modelling, extrapolate from your evaluation usage to production volume.
If the evaluation succeeds, your account team will move you to a paid Enterprise contract, at which point you can also enable SSO, request EU / US regional endpoints, and discuss on-prem deployment.
If the evaluation falls short, the team will help you understand why and (where applicable) suggest a fallback such as Pay-as-you-go for a narrower use case.
What both paths share
- Full access to Zero STT, Zero TTS, Vāķ Translate, and the intelligence layer, no feature gating during evaluation.
- Standard rate limits (16 concurrent requests on Pay-as-you-go; higher on Enterprise).
- Audio is encrypted in transit (TLS 1.3) and at rest (AES-256), and deleted on processing completion. See the security policy.
- Models are not trained on your audio or transcripts.