Your Simba voice, now answering the phone
Dograh is the self-hostable voice agent platform with an open-source workflow builder. Bring your Speechify key and a stock or cloned Simba voice, and wrap a phone agent around it that you host yourself or start on Dograh Cloud.
Key, model, voice, then a real call
Speechify sits in the voice settings
BYOKbring your own key
Add your SpeechifyAI key, and every reply the agent speaks runs through your own account, billed per character by Speechify.
→ A Reader plan does not include API usage
Simba 3.2 for English, 3.0 for five more
SimbaSpeechify's family of voice models
Simba 3.2 is the default and built for speed. Simba 3.0 speaks German, Spanish, French, Italian and Portuguese.
→ Match the model to the caller's language
Paste a cloned voice ID
Voice IDthe label for one stock or cloned voice
A clone made in Speechify works once it supports the Simba model you picked. Get the speaker's permission before you clone.
→ Test with a date, a price and a phone number
Speechify talks, you run the rest
Speechify still processes every line the agent speaks, under your own account
Clips play directly, so those lines add no per-character charge
Self-hosted concurrency begins at 10 calls and goes higher when you set it
Keep the sound, gain the agent
A voice that listens back
The agent hears the caller, works out the reply and speaks it in Simba, all inside one call flow.
Switch voices without a rebuild
Try Speechify, ElevenLabs or Kokoro on the same agent by changing only the voice provider.
Find the turn that went wrong
Each call's trace times speech, AI and voice per turn, and QA scores every step, so a misread price is easy to spot.
Open code around the voice
The whole agent platform is open under BSD-2-Clause, so your team can read and change it.
What a voice API leaves out
| What you get | Dograh | Speechify API |
|---|---|---|
| Listens to the caller and replies on a phone line | ✓ | ✕ |
| Books, looks up and transfers through tools mid-call | ✓ | ✕ |
| Open-source platform you can self-host | ✓ | ✕ |
Speechify, piece by piece
Which Speechify models can the agent use?+×
English calls default to Simba 3.2. Pick Simba 3.0 when callers speak German, Spanish, French, Italian or Portuguese.
Can I use my cloned Speechify voice?+×
Yes. Paste the cloned voice ID with your own key. The clone has to support the Simba model you picked.
Does call audio go to Speechify?+×
The agent's replies do, since Speechify turns them into speech under your key. Open-weight voices such as Kokoro keep that step on your servers.
Who pays for the Speechify usage?+×
Speechify bills your key. Self-hosted Dograh has no per-minute platform fee, but servers, model keys and phone minutes still cost money. Dograh Cloud charges a 1¢-a-minute platform fee.
Can the agent hand a caller to a person?+×
Yes, on Twilio, Telnyx, Plivo, Cloudonix and Asterisk ARI. Your colleague is rung first, and the caller is connected after a pickup.
Why test the voice on a real phone line?+×
A phone line carries 8 kHz audio against a browser's 16 kHz, so a demo favourite can sound thinner once it reaches a call.
Give your Simba voice its first phone call.
Self-host in one command, or start on Dograh Cloud.
$ curl … start_docker.sh && ./start_docker.sh