text-to-speech api built for real-time conversations

built to be versatile and human

type to insert emotion tags

built for natural, real-time voice applications

mulberry responds in 162ms, the fastest ttfb on the market, at ₹0.40/min (~$0.005/min).

1. low-latency speech

for voice agents, calls and conversational experiences.

2. design any voice

describe age, accent, pitch, tone and emotion instead of choosing only from a fixed catalogue.

3. multilingual conversations

switch naturally across hindi, english, hinglish and other languages.

4. production-friendly pricing

transparent price per minute, shown clearly against common alternatives.

implement silk in your app today

developer implementation is easy,….?

from rumikai import Rumik

client = Rumik()

# muga: expressive, steer with an inline tone tag
audio = client.speech.create(text="[happy] Namaste! Kaise hain aap?", model="muga")
audio.save("hello.wav")

# mulberry: faster, steer with a description + preset speaker
audio = client.speech.create(
    text="Hi there, how can I help you today?",
    model="mulberry",
    description="a warm 30s female voice, conversational pacing",
    speaker="speaker_2",
)
audio.save("greeting.wav")

silk is human in every language

hi there good news, your order just shipped. you should see it tomorrow.
hi there good news, your order just shipped. you should see it tomorrow.
hi there good news, your order just shipped. you should see it tomorrow.
hi there good news, your order just shipped. you should see it tomorrow.
hi there good news, your order just shipped. you should see it tomorrow.
hi there good news, your order just shipped. you should see it tomorrow.

pay for what you need

browse credit packs

silk mulberry

sota tts model

per 1k characters

₹0.50

silk muga

sota voice design

per 1k characters

₹0.99

silk spider

our most advanced model

contact the team for preview access

faqs

silk is rumik's expressive, multilingual text-to-speech api. send text, get natural, human-sounding speech back in real time.
silk mulberry starts at $0.005 per 1k tokens (₹0.40/min in india). pay for what you need with credit packs — no subscriptions.
english, hindi, hinglish, tamil, telugu, bengali, marathi and more — with natural mid-sentence switching between them.
yes. websocket streaming with low time-to-first-byte, fast enough for live conversations.
yes. silk is built for voice agents, calls and conversational experiences, with pipecat and livekit integrations.
describe the voice you want — age, accent, pitch, tone and emotion — instead of picking from a fixed catalogue.
credits are deducted per token generated. track usage anytime from the playground.
generation pauses until you top up. your account and api keys stay active.
limits depend on your plan. contact the team if you need higher concurrency for production workloads.
yes, commercial usage is allowed on paid plans.
a few lines of code. rest and websocket apis with python and node examples in the docs.

start building with rumik's tts api