01
BuildNo-Code Agent Creation
Build and configure voice agents without writing code — describe the conversation, set the languages, publish.
Features
Everything a public service needs to answer aphone call— and nothing it doesn't.
VoicEra turns a plain-language description into a working multilingual voice agent, connects it to the models and carriers you choose, and runs it on infrastructure you control.
- 23+
- Languages supported
- <2s
- Typical round-trip
- Open source
- MIT License
- Sovereign
- By design
Key capabilities
Built for real callers, on real phone lines.
02
BuildPlug-and-Play AI Models
Choose and combine integrated STT, TTS and LLM models per agent. Swap one layer without touching the rest.
03
BuildTest on Browser
Talk to an agent directly from the browser and refine it before it ever touches a phone line.
04
ConnectTelephony & Voice Integrations
Connect multiple telephony providers so the same agent answers on a real number.
05
In progressConnectWebSocket & SIP Integration
Real-time voice over WebSocket today; SIP integration is in progress.
06
BuildKnowledge Base
Ground agents in organisation-specific circulars, FAQs and registries so answers quote the published rule.
07
OperateCampaign Management
Create, manage and monitor outbound voice campaigns at scale — reminders, advisories, surveys.
08
OperateAnalytics & Telemetry
Track KPIs, call quality, usage and outcomes per agent, per language, per district.
09
OperateCall History & Transcripts
Access historical calls and their transcripts for audit, QA and continuous improvement.
10
LanguageLow-resource language support
Built for languages mainstream voice AI leaves out — beyond the 22 scheduled languages to Bhili, Gondi, Tulu and more.
11
LanguageDeterministic Language Switch
Move a live call to another language on a defined, predictable rule — not a guess.
12
LanguageAutomatic Language Detection
Recognise the language a caller is speaking and answer in it, without asking.
Ecosystem
Model partners and telephony you can plug in.
12 model providers across speech-to-text, reasoning and text-to-speech, and 4 telephony partners — pick sovereign, open or commercial per language, per agent.
IndiaKenpath
Indic-tuned LLM from the team behind MahaVistaar.
LLM
OpenAI
GPT models for reasoning, Whisper and TTS voices.
LLMSTTTTS
Qwen
Open-weight multilingual models, self-hostable.
LLM
Grok · xAI
xAI's Grok models over API.
LLM
Anthropic
Claude models for careful, grounded dialogue.
LLM
Groq
Low-latency inference for open models and Whisper.
LLMSTT
Azure OpenAI
Azure OpenAI and Azure Speech with India regions.
LLMSTTTTS
IndiaAI4Bharat
IndicConformer ASR and Indic TTS across 22 languages.
STTTTS
IndiaSarvam AI
Indic speech and language models built in India.
STTTTSLLM
ElevenLabs
Natural multilingual voices where quality is paramount.
TTSSTT
Deepgram
Streaming speech recognition and voices.
STTTTS
Cartesia
Ultra-low-latency streaming voices.
TTS
IndiaVobiz
Indian cloud telephony with virtual numbers and SIP trunks.
Telephony
Plivo
Programmable voice with India-local number inventory.
Telephony
Vodafone
Carrier connectivity for toll-free and enterprise lines.
Telephony
IndiaBSNL
Public-sector carrier reach across rural India.
Telephony- +
Custom model
Bring your own STT, LLM or TTS over any OpenAI-compatible or self-hosted endpoint.
LLMSTTTTS
Logos belong to their respective owners and indicate providers VoicEra can connect to; availability of a given model depends on your deployment and licence.
Custom model
Bring your own model.
Every layer is an adapter. Point VoicEra at any OpenAI-compatible endpoint, a self-hosted ASR or a fine-tuned voice inside your own network — including models that never leave a government data centre.
Custom LLM
+ Add Custom LLM- ProviderOpenAI-compatible
- Base URLhttps://llm.your-dept.gov.in/v1
- Modelindic-instruct-8b
- Data residencyOn-premise · Bengaluru
Connected · used by 3 agents
Build one yourself in a minute.
Describe an agent and talk to it in your browser, or open the Sandbox and take it further.