OSS

Custom Voice Stack Assembly (Self-Rolled)

Bespoke voice agent stack assembled from individual OSS components (Whisper, Llama, Coqui, Sarvam, custom).

Rs 2,99,999

Flat Fee

3-4 weeks

Delivery Timeline

10

Deliverables

GST

Invoice Included

Razorpay Verified

GST Invoice

Fixed Price

Scoping Call First

What you get

Custom Voice Stack Assembly (Self-Rolled)

For sophisticated buyers who don't want a framework like Pipecat or LiveKit — they want individual components stitched together for maximum control. Voice as a core product capability, not a feature bought from a vendor. We pick the right STT (Whisper / Sarvam / AI4Bharat / custom), LLM (Llama / Mistral / GPT-4 via Ollama / Anthropic / hybrid), TTS (Coqui / ElevenLabs / AI4Bharat / custom), telephony layer (Twilio / Plivo / Exotel / custom SIP), and orchestration approach (state machine, LLM-driven, or hybrid). Then we build the whole assembly in your repo, no framework dependency. Highest complexity, longest timeline, deepest customization in the catalog.

Deliverables

What's included

Every item below is delivered before final payment.

Component selection report with rationale

Custom voice stack code in your Git repo (no framework dependency)

STT integration (Whisper / Sarvam / custom)

LLM integration (local / cloud / hybrid)

TTS integration (Coqui / ElevenLabs / custom)

Telephony layer (Twilio / Plivo / Exotel / custom SIP)

Orchestration code (state machine or LLM-driven)

Monitoring + alerting

Deployment runbook + architecture documentation

60 days post-launch support

Scope & Responsibilities

What's in scope (and what isn't)

Honest framing of the engagement. Self-hosted deployments give you control — and the ongoing infra bills come with that control.

Edesy delivers
What we do
Included in the fixed fee for this package.
  • Architecture design and infrastructure provisioning runbook
  • Framework setup and configuration in your repo
  • Voice agent code, integration code, and webhook handlers
  • India language tuning (Hindi default; regional + niche on add-on)
  • Production hardening: error handling, retry logic, monitoring
  • Deployment + post-launch support window
You own
What stays on your side
Costs we don't pay or pass through. You retain accounts and pay the bills directly.
  • Cloud infrastructure account (AWS / GCP / Azure / your own)
  • LLM API account and bills (OpenAI / Anthropic / Gemini / etc.)
  • STT API account and bills (Deepgram / Sarvam / Azure / etc.)
  • TTS API account and bills (ElevenLabs / Cartesia / Sarvam / etc.)
  • Telephony account and bills (Twilio / Plivo / Exotel / etc.)
  • Ongoing infrastructure operations (covered if you add the maintenance retainer)
Want us to take infra and ops off your plate too? Add an OSS Voice Maintenance Retainer (Rs 29,999 - Rs 99,999/month depending on tier). Or move to the Edesy managed platform where the per-minute rate includes all the infra and ops by default.
Best For
Who buys this

Deep-tech teams building voice as a core product capability who reject framework abstractions

Pricing
Rs 2,99,999 flat
3-4 weeks

Plus 18% GST at Razorpay checkout. Add your GSTIN to claim Input Tax Credit. Per-minute platform usage billed separately on prepaid wallet at voice-agent.edesy.in.

Process

How this works

1
Inquire / Buy

Buy Now triggers Razorpay checkout. Inquire First books a scoping call.

2
Scoping call (30 min)

Confirm fit, deliverables, success criteria, go-live date. No payment yet for Inquire path.

3
Razorpay invoice / payment

Rs 2,99,999 + 18% GST. Add GSTIN to claim Input Tax Credit.

4
Delivery

3-4 weeks from kickoff. Weekly check-ins; daily near launch.

5
Sign-off + training

Acceptance review, knowledge transfer, runbook handover.

FAQ

Ready to start custom voice stack assembly (self-rolled)?

Buy directly via Razorpay (GST invoice included), or talk to sales first if you need a custom scope.