Models
Fourier STT listens, Gemma thinks and Fourier TTS speaks, on our GPUs in India.
Three models
The voice by the character, the ears by the hour, the model by the token.
Text to speech
₹1.80per 1,000 characters²
Pay as you go
Business
Speech to text
₹18an hour of recorded audio²
Language model
Gemma 4 26B A4B
Pay as you go
Business
Language model · Gemma 4 26B
You build the agent. We run the model, the capacity and the serving.
01
The 26B A4B architecture activates roughly 4B parameters per token. A mixture of experts, built to use compute selectively.
02
We tune precision, batching and decoding around live-call workloads, with first-token latency and sustained throughput in view.
03
Streaming text and tool calling let the agent respond, look something up or trigger the next step in a conversation.
It runs inside every Frequency agent today. To call it on its own, talk to us and we switch it on for your workspace.
Languages
Hinglish too, the mix in between. Bengali, Marathi and Kannada come next.
One line each, straight from Fourier TTS.
Text128 characters
नमस्ते, मैं नेहा बोल रही हूँ। आपका पार्सल कल शाम तक पहुँच जाएगा। अगर उस समय आप घर पर नहीं होंगे, तो बता दीजिए, हम दिन बदल देंगे।
48 kHz · MonoReady when you are
0:00 / 0:09
Fourier TTS, voice neha, rendered 16 Oct 2026. One request, not edited.
Text128 characters
Hello, मैं समीर, Acme Motors service centre से। आपकी bike की free service इस month due है। Slot book कर दूँ, Saturday या Monday?
48 kHz · MonoReady when you are
0:00 / 0:08
Fourier TTS, voice sameer, rendered 16 Oct 2026. One request, not edited.
Text158 characters
Hi, I am Shruti from Acme Dental Clinic. Your appointment is tomorrow at eleven thirty in the morning. Shall I confirm it, or would you like a different time?
48 kHz · MonoReady when you are
0:00 / 0:08
Fourier TTS, voice shruti, rendered 16 Oct 2026. One request, not edited.
Text117 characters
નમસ્તે, હું આશુ બોલું છું. તમારી લોનનો હપ્તો આવતા અઠવાડિયે ભરવાનો છે. તમે કહો તો હું અત્યારે જ ભરવાની રીત સમજાવી દઉં.
48 kHz · MonoReady when you are
0:00 / 0:08
Fourier TTS, voice ashu, rendered 17 Oct 2026. One request, not edited.
Text167 characters
வணக்கம், நான் நேஹா பேசுகிறேன். உங்கள் பார்சல் நாளை மாலைக்குள் வந்து சேரும். அந்த நேரத்தில் நீங்கள் வீட்டில் இல்லையென்றால் சொல்லுங்கள், வேறு நாளுக்கு மாற்றி விடுகிறோம்.
48 kHz · MonoReady when you are
0:00 / 0:10
Fourier TTS, voice neha, rendered 17 Oct 2026. One request, not edited.
Text114 characters
నమస్కారం, నేను సమీర్ మాట్లాడుతున్నాను. నిన్న మీరు మా కొత్త ఇళ్ళ గురించి అడిగారు. ఇప్పుడు కొంచెం సేపు మాట్లాడవచ్చా?
48 kHz · MonoReady when you are
0:00 / 0:08
Fourier TTS, voice sameer, rendered 17 Oct 2026. One request, not edited.
How we ship
Three things that happened, and the numbers that made us act.
38.4%, per-turn pitch spread, against 7.6% shipped
A new voice release sounded fine on a single line. Pitch analysis measured the speaker drifting turn to turn — 38.4% spread against the shipped model’s 7.6%. We rolled back the same hour.
Read the engineering note~6 hours, dark, no calls dropped
A GPU provider’s API and console went unreadable. Our fleet controller treats “cloud unreadable” as “leave the node alone”, not “tear it down”, so production kept serving. Customers never knew.
Read the engineering noteNothing is hand-edited on a live box. What runs is always a commit that exists in history, so “fixed locally, not on the box” stopped being a failure mode we can have.
Read the engineering noteQuestions
Hindi, English, Gujarati, Tamil and Telugu are live in Fourier TTS and Fourier STT, and so is Hinglish. Fourier TTS also code-switches between Devanagari Hindi and Roman English in one sentence. Bengali, Marathi and Kannada come next.
₹1.80 per 1,000 characters from a prepaid wallet. Pay as you go has 5 concurrent requests; Business has 25, with a ₹50,000 monthly commitment. All prices exclude GST.
₹18 an hour of recorded audio, charged by the second once a job completes. Degraded and failed jobs are free, and a recording can run up to 2 hours. All prices exclude GST.
On request. Gemma 4 26B runs inside every Frequency agent, and on its own it is priced by the token. Talk to us and we switch it on for your workspace.
Fourier TTS speaks OpenAI’s POST /v1/audio/speech protocol, so the OpenAI SDKs work with a new base URL and model name. The same key works for the rest of the v2 API.
Fifteen minutes with the team that builds it. Bring your script, or your hardest call.
Book 15 minutes