Skip to content
We built the world's most realistic voice model. #1 on Design Arena
Book a demo

Voice AI for
banking and insurance.

Turkish-speaking AI for banking and insurance.

Request a demo
Illustrative conversations: loan tracking, missing claim documents and a promise to pay.

Working with teams in banking and insurance.

Across other industries
Doğuş Oto
Enterprise Rent-A-Car
ikas

Partnership programs we participate in

HackZone by Allianz
Garanti BBVA Partners Tech
Mastercard Lighthouse Türkiye
“We tried models from many different providers but couldn't achieve the sound quality we were looking for. Freya was the first solution that met our expectations. Thanks to their team that took quick action, the installation process proceeded extremely smoothly.”
Jiri LahodaSuri.czChief Executive Officer
“In our proof-of-concept work with Freya, we were able to try different insurance scenarios very quickly. The team's flexible, solution-oriented approach made the process easier for us.”
Aydın BozdemirAnadolu SigortaDeputy Chief ExecutiveTranslated from Turkish.
“Freya helped us create a 24/7 voice operation for our embedded BNPL product without establishing a call centre. Their assistants manage outbound reminders and inbound frequently asked questions comprehensively, adhere precisely to our policies, and integrate directly into our workflows.”
Süleyman OnayColendiManaging Director
“Freya, whom we have worked with for over a year, has developed impressive AI technology that delivers natural, seamless conversations.”
Ömercan ÇeliklerikasCo-FounderTranslated from Turkish.

News · Design Arena, Audio Realism Bench

We built the world's most realistic voice model.

Freya TTS Adam V1 ranks first among models on Design Arena's blind listening test, just behind human speech. Listeners hear two clips and pick the one that sounds more human. The ranking is an Elo rating.

Listen to Adam V1Public leaderboard
Audio Realism BenchDesignArenaby Intelligence
  1. 1462
    Human speech
  2. 1419
    Freya TTS Adam V1
  3. 1369
    Bland Speech v3
  4. 1306
    Cartesia Sonic 3.6
  5. 1202
    Eleven v3 Conversational
  6. 1188
    Cartesia Sonic 3.5
  7. 1150
    Grok TTS
  8. 1101
    Google Gemini 2.5 Pro TTS
  9. 1071
    OpenAI GPT Realtime 2
Elo rating, blind listening test. Selected models. Measured 2026-09-16.

A better experience
for banking and insurance customers.

From your customer's question to the next action in your systems.

Automate your banking operations

Illustrative scenario

Loan application follow-up

Request
Application status
Source
Bank application system
Next step
Notify about a missing document
Freya

We need your proof of income to review your application. You can upload it through mobile banking.

Conversation outcome recorded

Send payment reminders, follow up on loan applications and help customers complete missing documents using your approved information.

Turkish that sounds natural

Turn text into speech and speech into text with models trained by Freya.

Explore the APIs

Connected to your systems

CRMAPIFreyaTelephonyWorkflow

Connect your CRM, telephony and workflows. Give customers up-to-date answers.

Explore integrations

Test before you go live

Identity verificationAction permissionsAgent handoff

Simulate real-world scenarios. Review issues, refine your workflow and test again.

Explore simulations

Add Text to Speech to your app in minutes

from openai import OpenAI

client = OpenAI(
    base_url="https://tts.freyavoice.ai/v1",
    api_key="FREYA_API_KEY",
)

speech = client.audio.speech.create(
    model="freya-tts",
    voice="leyla",
    input="Merhaba, ben Freya. Size nasıl yardımcı olabilirim?",
    response_format="wav",
)
speech.write_to_file("speech.wav")
Request API access

Built around the way you work.

From deployment to daily operations, you stay in control.

01

Your deployment choice

On-prem or cloud hosted in Türkiye. We work with you to choose a setup that meets your data and access requirements.

02

Your rules and permissions

Define what information your agent can access, which actions it can take and when it should hand over to a person.

03

Conversations you can inspect

Review recordings, transcripts and action logs. Understand what happened and evaluate outcomes with your team.

04

A team through go-live

Our engineers work with your team on process design, integration and scenario testing. On-premises updates are rolled out with your team.

Security and deployment options

Freya Research

We build our own Turkish speech models.

We work on speech synthesis, speech recognition and voice interaction. With FreyaTTS-small, we have released our code, model weights and evaluation dataset for anyone to use.

Explore the research
Freya Research2026

Turkish speech synthesis

FreyaTTS-small

Our compact model, trained from scratch to turn Turkish text into speech. Read the paper, inspect the results and run it on your own infrastructure.

Audio output
48 kHz
License
Apache 2.0
Paper
Explore in 3DModelCodeBenchmark

Help build the next model.

Working on speech or language models? We'd like to meet you. Email us your research, code or the questions you want to explore.

Apply to the research team

We also support emerging researchers. Discover the YZ50 program

Research and Updates

All updates
FreyaTTSPaper · 2026

FreyaTTS paper on arXiv

Our compact tokenizer-free flow-matching transformer for Turkish-first speech synthesis outperforms much larger models at 40-55% of their size and runs in real time on consumer hardware.

See what Freya can do for you.

Tell us what you want to automate. We'll explore Freya using a scenario from your business.

Book a demo
Freya | Voice AI for Banking and Insurance