Products
AI Voice Agent Platform
Build LLM-powered AI Voice Agents
Explore more →

MRCP ASR
ASR for VoiceXML IVR Platforms
Explore more →
Developers
APIs, SDKs, Docs & Pricing
Explore more →
By Industry
Help
Documentation
↗Implementation guides
Status Page
↗System health & uptime
Support
↗Get help from the team
Meet Voicegain MRCP ASR. A future-proof neural-network STT that talks MRCP and supports both speech grammars and large vocabulary STT. Keep your IVR for now and migrate to Conversational AI on your timeline.
Trusted by companies building amazing voice AI products





01
Native support for SRGS, grXML, JSGF, and built-in grammars. MRCPv1 and v2. Your VoiceXML application doesn't change a line.
02
Deploy using Docker Compose. Just point the ASR IP to Voicegain. CPU & GPU based VMs. K8s for large scale deployments.
03
Based on number of Ports. Call-home or Air-gapped License server. Tier 1 - Grammar based ASR. Tier 2 - Large Vocabulary STT.
Voicegain MRCP is a standards-compliant MRCPv1 and MRCPv2 server, packaged with our deep-learning based ASR/STT engine. If your platform speaks MRCP, it can speak Voicegain.
Genesys
Avaya
Cisco
Aspect(Voxeo)
Syntellect
FreeSWITCH
Asterisk
Dialogic
92%
In Grammar accuracy
12,500
Ports of MRCP ASR
5
Number of VoiceXML Platforms
Don't see your platform? Talk to us →
Legacy MRCP ASR
An aging ASR that is approaching EOL.
Grammar-based ASR has now reached EOL
Legacy Vendors are no longer maintaining their current grammar-based ASR offerings. EOL in 2027/28.
No easy upgrade path
New ASR requires gRPC as the way to access - a protocol that On-premise VXML platforms do not support
Cloud-only option
MRCP gateways from STT Vendors require you to send sensitive audio beyond the firewall, breaking compliance
Decade-old acoustic models
Accuracy of old grammar-based STT models is very poor. It has not kept pace with modern large vocabulary STT engines.
Rip-and-replace pressure
Vendors are pushing you to replace the entire IVR stack and ASR- entirely instead of just the speech layer
Voicegain MRCP
A Modern STT that supports old and new.
Support both grammar-based and Large Vocabulary
Our modern STT models natively support both grammar-based and large voocabulary STT.
Drop-in replacement with MRCPv1 or v2
Your VoiceXML IVR app doesn't need to change a line of code. All you need to do is to point the ASR resource to the new IP address.
Deploy in Your VPC or Datacenter
Same VXML Platform, Same IVR Application. Your network end-to-end. No forced cloud migration.
Deep-learning STT, trained on call audio
89% broad benchmark accuracy. Hundreds of thousands of hours of training data. 1B+ minutes transcribed in production.
Path forward, not rip-and-replace
Upgrade your speech engine now with a simple docker compose deplpyment. Modernize to conversational AI on your timelines.
The same Voicegain ASR that can support a grammar-based app also supports Conversational AI Agents. You're not buying a dead-end product.
Today
Drop in replace Voicegain MRCP ASR. Keep your existing speech IVR with no changes to App logic. Extends the life of your IVR.
When you're ready
Use the same model for conversational interactions where grammars don't fit. Real-time captioning, voice bots, call analytics.
Eventually
Use the Voicegain AI Voice Agent Platform. Bring your own LLM. Same network, same STT models, on your timeline.
Match your procurement model. Both options ship with our deep-learning ASR included.
Standard
Regulated & secure
Per-port licensing with no external license callbacks. For datacenters with no outbound internet — government, healthcare, finance.
Is Voicegain a drop-in replacement for Nuance Recognizer?
+Yes — point your IVR's ASR resource at the Voicegain MRCP server instead of Nuance. Your application, call flow, and grammars stay exactly as they are.
Can Voicegain MRCP run fully on-premises?
+Yes — Voicegain MRCP supports call-home or fully air-gapped licensing for datacenters with no outbound internet connectivity.
What grammar formats do you support?
+Native support for SRGS, grXML, JSGF, and built-in grammars, across both MRCPv1 and MRCPv2.
What languages do you support?
+Voicegain's ASR supports a broad set of languages, with deep-learning models trained on hundreds of thousands of hours of call audio.
What happens when we're ready to move beyond grammar-based IVR?
+The same Voicegain ASR that powers MRCP also supports gRPC, WebSockets, and large-vocabulary streaming — and connects directly to the Voicegain AI Voice Agent Platform when you're ready for Gen AI voice bots.
Full compliance with healthcare regulations.
Audit complete on all 5 criteria of SOC 2.
Complies with Level-2 v4.0 PCI guidelines.
Secure solutions from call start to finish.