English · Male
Varun
Built-in English speaker available through the Radiance API and playground.
Radiance · English
Turn English text into natural, expressive speech for voice agents, customer conversations, applications, narration, and real-time AI experiences. Choose a voice, control the delivery, and generate spoken English through Shunya Radiance.
Hello, thanks for calling. I can help you check your account, update your details, or answer a question about your recent transaction.
Conversational delivery · RadianceEnglish TTS
Text-to-speech is not only about converting words into audio. Production speech needs the right voice, pacing, expression, output format, and response time for the application using it.
46
Built-in speaker voices available across the Radiance voice catalogue
11
Expression styles for changing how speech is delivered
<350ms
First audio in streaming mode for real-time applications
7
Audio output formats for web, production, streaming, and telephony
| Capability | Radiance · English Text to Speech |
|---|---|
| English Male Voice | Varun |
| English Female Voice | Nisha |
| Expression Styles | 11 |
| Batch Synthesis | ✓ |
| Streaming Synthesis | ✓ |
| Cross-lingual Synthesis | ✓ |
| Voice Cloning | ✓ |
| Speed Control | 0.25× - 4× |
| Audio Formats | 7 |
Try it yourself
The best way to evaluate a voice is with the words your application will actually say.
English voices
Radiance includes two named English voices in its standard voice catalogue.
English · Male
Built-in English speaker available through the Radiance API and playground.
English · Female
Built-in English speaker available through the Radiance API and playground.
But the voice catalogue does not stop at two English speakers. Radiance separates voice identity from the language being spoken. Any supported voice can synthesize supported languages, enabling teams to experiment with different vocal characteristics without changing the input language.
46 voices
One voice catalogue. Cross-lingual synthesis built in.
| Language | Male voice | Female voice |
|---|---|---|
| English | Varun | Nisha |
| Hindi | Rajesh | Sunita |
| Bengali | Arjun | Priyanka |
| Tamil | Murugan | Thangam |
| Telugu | Vishnu | Lakshmi |
| Kannada | Kiran | Shreya |
| Marathi | Siddharth | Ananya |
| Malayalam | Krishnan | Deepa |
| Punjabi | Gurpreet | Simran |
| Urdu | Salman | Fatima |
The input text controls the language. The selected voice controls the vocal character.
Expressive speech
Real conversations do not use one tone for everything. A support response, an announcement, an educational explanation,
and a story may use the same language while requiring completely different delivery. Radiance supports
11 expression styles for controlling how generated speech sounds.
Natural everyday delivery for voice agents, assistants, and customer interactions.
Long-form delivery for stories, articles, explanations, and narration.
Structured delivery for announcements and informational content.
Brighter delivery for positive interactions and messages.
Higher-energy speech for engaging experiences.
Softer delivery for situations where the tone needs to reflect the content.
And moreExpression can be selected according to the context instead of forcing every sentence through the same speaking style.
Real-time speech
Pre-generated audio can wait. A voice agent cannot. Radiance supports both streaming and batch synthesis,using the same speech platform for two different production requirements.
<350ms
First audio in streaming mode
Streaming
Open a persistent WebSocket connection and receive audio progressively as synthesis takes place.
Best suited for:
Batch
Send the complete English text and receive the finished audio output.
Best suited for:
| Requirement | Streaming | Batch |
|---|---|---|
| Real-time playback | ✓ | - |
| Complete file output | - | ✓ |
| Voice agents | ✓ | - |
| Narration | ✓ | ✓ |
| WebSocket | ✓ | - |
| HTTP API | - | ✓ |
Audio control
The same audio format is not right for every application. Web playback, production audio, low-latency pipelines,
archival workflows, and telephony have different requirements. Radiance supports seven output formats.
| Format | Common use |
|---|---|
| MP3 | Web playback and general applications |
| WAV | High-quality audio workflows |
| PCM | Real-time and low-latency pipelines |
| OGG Opus | Efficient streaming and web audio |
| FLAC | Lossless audio |
| mu-law | Telephony systems |
| A-law | Telephony systems |
Speed control
Adjust speech speed from 0.25× to 4.0×. Applications can control delivery speed without creating a separate voice model.
Silence control
Speech output can also be configured to trim silence when tighter audio is required.
Voice cloning
Static voice recordings only work for sentences that have already been recorded. Voice cloning creates a reusable synthetic voice from a short reference sample.
3-6 seconds
Reference audio required for Radiance voice cloning
01
Supply a clean reference recording.
02
The reference is used to establish the voice characteristics for synthesis.
03
Generate English speech that was never present in the original recording.
04
Use the voice across supported languages instead of building an entirely separate voice identity for every language.
Built for consistency: One voice identity can remain part of the product as the words - and potentially the language - change.
Real-world use cases
Turn dynamically generated responses into spoken English during live AI conversations.
Create automated spoken responses for account questions, service requests, updates, and self-service workflows.
Generate dynamic phone prompts while supporting telephony-ready mu-law and A-law audio.
Convert lessons, explanations, instructions, and educational material into spoken English.
Generate audio for articles, stories, scripts, presentations, and other content.
Offer spoken alternatives to written interfaces and digital information.
Generate account alerts, reminders, order updates, appointment messages, and other dynamic speech.
Add a voice layer to assistants, copilots, conversational products, and other generative AI experiences.
Developers
Radiance can be integrated through HTTP for batch synthesis or WebSocket for streaming speech.
{
"model": "radiance-indic",
"input": "Hello, how are you today?",
"voice": "Varun",
"language": "en",
"response_format": "mp3"
}Input
From your application, content system, user workflow, or LLM.
Generate
Create complete audio through batch synthesis or receive progressive audio through streaming.
Configure
Output
Built for production
Generating one good sentence is not the same as operating a speech product. Production teams also need streaming, audiocontrols, APIs, multilingual expansion, deployment flexibility, and consistent voice behaviour.
| Requirement | Shunya Radiance |
|---|---|
| English Text to Speech | ✓ |
| Male + Female English Voices | ✓ |
| Expression Control | ✓ |
| Streaming TTS | ✓ |
| Batch TTS | ✓ |
| Cross-lingual Synthesis | ✓ |
| Voice Cloning | ✓ |
| Multiple Audio Formats | ✓ |
| Telephony Audio | ✓ |
| API Integration | ✓ |
| LLM → TTS Workflows | ✓ |
LLM → TTS
This makes English TTS part of a complete voice interaction rather than a standalone audio-generation step.
Questions
English Text to Speech converts written English into spoken audio using a speech synthesis model.
Radiance includes Varun, a male English voice, and Nisha, a female English voice, in its named voice catalogue.
Yes. Radiance supports streaming synthesis, designed for applications such as voice agents and real-time conversational experiences.
Yes. Radiance supports 11 expression styles for changing how generated speech is delivered.
Yes. Speech speed can be configured from 0.25× to 4.0×.
Radiance supports PCM, WAV, MP3, OGG Opus, FLAC, mu-law, and A-law.
Radiance supports voice cloning using a short reference audio sample.
Yes. Radiance supports cross-lingual synthesis across its supported language set.
Get Started
Don't judge a voice on our sample. Give it your text. Try the names, numbers, product terms, customer responses,scripts, or AI-generated sentences your application actually needs to speak.Choose a voice. Change the expression. Hear the result.