Deployment choice
Choose managed cloud, customer VPC or on-premise deployment.

Use one model family across managed cloud, customer VPC or on-premise deployments.
Choose managed cloud, customer VPC or on-premise deployment.
Run workloads inside customer-controlled infrastructure when required.
Use Perigee-1 speech and language interfaces across the selected architecture.
Deploy for the languages, accents and code-switching patterns your users actually use.
Put the Perigee-1 family to work across voice and language workflows. Choose the capabilities and deployment architecture that fit your team.
Transcribe calls, redact sensitive details and turn conversations into structured summaries for support teams.
Make multilingual patient conversations easier to document and review while keeping deployment requirements in view.
Turn customer voice interactions into searchable text and protect sensitive account details in downstream workflows.
Create transcripts and subtitles for recordings, then translate content for audiences across languages.
Make lessons and spoken material accessible as text, translated content and generated speech.
Support multilingual voice and text journeys for teams serving people across regions and languages.
Select the operating boundary that fits your workload. The final topology, capacity and responsibility model are confirmed during architecture review.
Run through OrbitalsAI's managed production infrastructure.
Suitable for
Conceptual topology
Final network boundaries, compute requirements and operating responsibilities are agreed during architecture review.
Verified controls and compliance status are stated plainly. Private networking, retention and residency requirements are scoped during deployment review.
Transport
Encryption in transit
Storage
Encryption at rest
Authentication
Bearer sessions and API keys
API keys
Hashed storage and revocation
Access
User-scoped jobs and results
Retrieval
Time-limited signed audio links
Published numbers are benchmarks, not contractual SLAs. Representative workload testing remains part of an enterprise evaluation.
~200 ms
Realtime end-to-end latency
Asynchronous batch processing
WebSocket streaming
Realtime word timestamps
Recorded transcript and SRT output
Speech, realtime, text and voice coverage are sourced from the shared product capability configuration.
Arabic · English · French · Hausa · Igbo · Kinyarwanda · Swahili · Twi · Yoruba · Zulu
Afrikaans · Amharic · Arabic · English · French · German · Hausa · Igbo · isiXhosa · Italian · Kinyarwanda · Sesotho · Setswana · Shona · Spanish · Swahili · Twi · Yoruba · Zulu
Use REST APIs, WebSockets and the verified Python SDK. The platform also provides API-key controls and visibility into requests, usage, processing and realtime sessions.
Read the documentation →import os
import orbitalsai
client = orbitalsai.Client(
api_key=os.environ["ORBITALSAI_API_KEY"]
)
transcript = client.transcribe(
file_path="call.mp3",
language="hausa",
model_name="Perigee-1-transcribe",
)
print(transcript.text)Your systems
OrbitalsAI
Internal applications
Tell us about your workload, languages and deployment requirements. We’ll work with you to evaluate the right architecture.