Progress Consulting
MotorOwn voice engine

Sign in

Test and tune Ana on our own voice engine.

Progress Consulting
MotorOwn voice engine · test console
Signed in as
Ana has unsaved changes. Tests use the last saved version.

Agent

Who answers in the tests below. Models, voice and all other settings are shared.

Off after every page load. Needs the test PC with its voice server and tunnel running; if it is not reachable the test uses the normal voice. Queue calls never use it.

Case data

What Ana says in this test. Saved as the default for the next test.

Talk in the browser

Uses your microphone. No phone line or Dinstar needed, good for the first test.

Call a phone

Through the Dinstar, like the real calls. Only numbers on the allowed list.

Chat by text

Type to Ana as the customer. Same prompt, AI model, masking and end-call tool as a call, without speech-to-text. Without "Speak replies" it only costs AI tokens.

Current test

Avg reply time
–
Replies
0

Test calls

The last 100. Click one for the transcript and latency per reply.

WhenTypeNumberStatusN / P / UDurationAvg replyCostPer minRec

Demo calls

Phone calls made from demo.progressconsulting.com.py, by login. Click one for the recording and transcript. Recordings are kept as long as the recording retention under Agents → Ana.

WhenLoginAgentNumberStatusTalkRec

Real usage

What the providers themselves counted, read with your keys. Speech-to-text needs the "Deepgram usage" key and OpenAI dollars the "OpenAI admin" key (API keys tab). Phone minutes are only on your carrier's invoice.

Loading…

Conversation

Copied from Ana in VAPI. Variables: {{customerName}}, {{productName}}, {{priceDetails}}, {{locationName}}.

Charla libre

The free-conversation agent (choose it on the Test tab). Same models, voice and settings as Ana; only these two texts differ.

Models and voice

Pick from the lists. "Other…" lets you type any model the provider offers.

0 = always the same wording, 1 = more varied.
1 = normal.
0 = unchanged, negative = quieter.

Turn taking

When Ana decides the caller has finished, and when she stops if interrupted.

0 = no limit.
Nobody speaks. 0 = off, minimum 15.

Phone line

Inverfin call queue

The queue can send calls here the same way it sends them to VAPI. Only the ports listed here are offered to it.

e.g. 28-30; empty = off.

Costs (USD)

Used for the cost shown on every call. Leave a price empty to use the list price of the model you selected (shown in grey).

ElevenLabs depends on your plan: monthly price divided by the characters it includes, times 1,000. Only per-use costs are counted: the phone line and the server are flat costs, and LiveKit runs on our server.

All changes saved

On this page

  • Conversation
  • Charla libre
  • Models and voice
  • Turn taking
  • Phone line
  • Inverfin call queue
  • Costs

Versions

Every save is kept. Open one to see what changed and restore it.

    API keys

    Each key is encrypted (AES-256-GCM) before it's saved on the server and is never shown again, only its last 4 characters. To change one, paste a new key over it.

    Minimum to test: OpenAI (model and voice) and Deepgram (speech-to-text). ElevenLabs or Cartesia only if you want those voices.

    Account

    Signed in as .