Kanji
漢字
Characters used extensively for nouns, verb roots, names, and many core concepts.
Radiance · Japanese
Convert Japanese text into natural AI speech for voice agents, customer experiences, narration, applications, media, learning, and real-time conversational AI. Use Shunya Radiance to generate spoken Japanese from native Japanese text, with streaming and batch synthesis, expressive delivery, configurable speed, and production-ready audio formats.
こんにちは。ご注文の商品は明日の午前中に到着する予定です。
Hello. Your order is expected to arrive tomorrow morning.Japanese TTS
Japanese speech synthesis needs to do more than read characters aloud. Written Japanese combines multiple scripts, context-dependent readings, numbers, names, punctuation, English terminology, and different levels of conversational formality. Radiance includes Japanese in Shunya's extended world-language speech-synthesis coverage.
11
Languages on Radiance Oriental
11
Expression styles available across Radiance
<350ms
First-audio streaming latency
7
Supported audio output formats
Japanese is part of global TTS: Shunya's named 46-voice catalogue is centred on Indic languages and English. Japanese uses Radiance Oriental, which has its own native Japanese voices, Hina and Haruto.
| Capability | Radiance · Japanese Text to Speech |
|---|---|
| Native Japanese Text | ✓ |
| Streaming Synthesis | ✓ |
| Batch Synthesis | ✓ |
| Expression Control | ✓ |
| Speed Control | ✓ |
| Multiple Audio Formats | ✓ |
| Web Audio | ✓ |
| Telephony Audio | ✓ |
| API Integration | ✓ |
| Voice Agent Integration | ✓ |
| Cross-lingual Speech Platform | ✓ |
Try it yourself
Don't evaluate Japanese TTS only with a prepared sentence. Try a customer response, product name,address, notification, dialogue, instruction, narration, or Japanese-English sentence.
Japanese writing system
Japanese text commonly combines three writing systems inside the same sentence.
Kanji
Characters used extensively for nouns, verb roots, names, and many core concepts.
Hiragana
Used for grammatical endings, particles, native words, and other language functions.
Katakana
Frequently used for foreign words, brand terminology, technical vocabulary, and emphasis.
新しいアプリをダウンロードしてください。
This single sentence contains:
A Japanese TTS workflow needs to process the complete sentence rather than treating each script as a separate language.
Context changes pronunciation
Japanese contains characters whose reading can depend on the word and surrounding context. Names, numbers, counters, abbreviations,and imported words introduce more variation. The goal is understandable Japanese speech - not character-by-character audio.
Readings that depend on the word and surrounding context.
Particles, endings, and native grammatical structure.
Foreign words, brands, and technical terms.
People, places, and companies with pronunciation variation.
Counters, dates, and numeric expressions.
Calendar and clock expressions in running text.
Short forms and imported acronyms.
English and Latin-script terms inside Japanese sentences.
Sentence boundaries and emphasis that shape delivery.
Natural Japanese delivery
Japanese speech changes according to context. A customer-service response should not necessarily sound like news narration. Radiance provides expression controls for changing the delivery of generated speech - 11 styles across the platform.
General-purpose speech.
Natural everyday interactions and voice agents.
Long-form narration and storytelling.
Formal informational delivery.
Upbeat speech.
Softer, subdued delivery.
More energetic speech.
More intense speech.
Anxious or tense delivery.
Same Japanese text: ご利用いただきありがとうございます。 - Neutral, Conversational, or Enthusiastic. The text remains the same. The delivery changes.
Japanese + English text
Japanese digital communication regularly contains English in product, software, and customer-facing content.
Shunya APIを使ってリアルタイム音声を生成できます。
Premiumプランへのupgradeが完了しました。
Customer Supportチームにお問い合わせください。
Modern applications should expect Japanese text to contain Latin characters and English terminology.
Mixed-script text should not require applications to manually create a separate audio file for every English term.
Real-time Japanese TTS
Pre-generated speech works for fixed announcements. Voice agents need dynamic speech.Radiance supports streaming synthesis through WebSocket connections.
<350ms
First-audio latency
Stream instead of waiting - the complete paragraph does not need to finish generating before the speech pipeline begins producing audio.
Streaming
Use WebSocket TTS when speech needs to begin quickly.
Best for:
Batch
Generate a complete audio file from finished Japanese text.
Best for:
| Requirement | Streaming | Batch |
|---|---|---|
| Voice agents | ✓ | - |
| Live conversation | ✓ | - |
| Complete audio file | - | ✓ |
| Narration | ✓ | ✓ |
| WebSocket | ✓ | - |
| HTTP | - | ✓ |
Control the Japanese audio
Different applications need different audio formats. Radiance supports seven output formats.
| Format | Best for |
|---|---|
| PCM | Real-time pipelines |
| WAV | Editing and high-quality files |
| MP3 | Web and general delivery |
| OGG Opus | Web streaming |
| FLAC | Archival and lossless workflows |
| μ-law | PSTN / telephony |
| A-law | SIP and compatible telephony |
Speaking speed
Control speech speed for web, voice agents, and media workflows.
Trim silence
Trim silence around generated audio where required.
One TTS API. Different delivery environments.
Japanese voices
Japanese is synthesized with Radiance Oriental, Shunya's text-to-speech model for Asian languages.It includes two native Japanese voices, Hina and Haruto. Choose one by namein the voice field of any batch or streaming request.
2
Native Japanese voices: Hina and Haruto
Model
Set the model value to radiance-oriental and the language value to ja.
Female voice
A native Japanese voice for voice agents, customer support, notifications, and narration.
Male voice
A native Japanese voice for voice agents, announcements, product audio, and narration.
Choose a voice
Pass Hina or Haruto as the voice value. The same voice works in batch and streaming requests.
Batch and streaming: Hina and Haruto work in HTTP batch and WebSocket streaming requests, with the same expression, speed, and output format controls.
Cross-lingual voice experience
Shunya's current public Radiance experience demonstrates a multilingual voice flow where thesame voice identity can move across languages - including English, Hindi, Japanese, Tamil,and Korean. The language changes. The intended voice identity remains consistent.
Multilingual applications can use one recognisable voice experience across different markets. For production voice-cloning availability and configuration, use the capabilities enabled for the specific account or deployment.
Japanese voice agents
Text-to-speech becomes particularly important inside voice-agent systems.
注文状況を確認したいです。
Converts speech into text.
Understands the request and generates a Japanese response.
Generates spoken Japanese.
ご注文の商品は現在配送中です。
STT → LLM → TTS
Real-world use cases
Generate dynamic Japanese responses for conversational AI.
Provide spoken Japanese support experiences and automated responses.
Build Japanese-speaking kiosks, product assistants, and customer interfaces.
Generate spoken instructions and assistance for Japanese-speaking guests and travellers.
Turn appointment updates, order messages, reminders, and alerts into Japanese speech.
Generate Japanese audio for lessons, exercises, educational material, and pronunciation experiences.
Create Japanese narration and speech for digital content.
Provide spoken alternatives to Japanese written interfaces.
Generate dynamic Japanese dialogue for supported applications.
Add Japanese speech output to existing AI and business workflows.
Developers
Radiance provides HTTP batch and WebSocket streaming synthesis. Batch: POST https://ttsv2.shunyalabs.ai/v1/audio/speech · Streaming: wss://ttsv2.shunyalabs.ai/v1/realtime. Standard ISO language code: ja.
{
"model": "radiance-oriental",
"language": "ja",
"input": "<Conversational> こんにちは。今日はどのようなお手伝いができますか?",
"voice": "Hina",
"response_format": "mp3"
}Basic workflow
Configure
Japanese requests use Radiance Oriental: set model to radiance-oriental, language to ja, and voice to Hina or Haruto, the two native Japanese voices.
Japanese for production
Generating one Japanese sentence is easy to demonstrate. Production speech systems need real-time synthesis, APIs, audio-formatcontrol, predictable latency, language handling, speech styles, telephony support, and deployment flexibility.
| Requirement | Radiance |
|---|---|
| Japanese Text to Speech | ✓ |
| Japanese World-language Support | ✓ |
| Native Japanese Text Input | ✓ |
| Kanji / Kana Input | ✓ |
| Batch Generation | ✓ |
| Streaming Generation | ✓ |
| First Audio | <350ms |
| Expression Styles | 11 |
| Speed Control | 0.25×-4.0× |
| Audio Formats | 7 |
| MP3 / WAV | ✓ |
| PCM / OGG Opus | ✓ |
| FLAC | ✓ |
| Telephony Formats | ✓ |
| API Integration | ✓ |
| LLM → TTS Workflow | ✓ |
| Voice Agent Integration | ✓ |
| Enterprise Deployment | ✓ |
Questions
Japanese Text to Speech converts written Japanese into spoken audio using speech-synthesis technology.
Yes. Japanese is synthesized by Radiance Oriental, Shunya's text-to-speech model for Asian languages.
No. Japanese is synthesized by Radiance Oriental, not the Indic-language catalogue.
Japanese TTS is designed to synthesize Japanese text, including normal Japanese writing that combines Kanji, Hiragana, and Katakana.
Yes. Radiance Oriental includes two native Japanese voices: Hina (female) and Haruto (male). Choose one by passing its name as the voice value in a batch or streaming request.
Yes. Radiance supports WebSocket streaming with first audio in under 350 ms.
Yes. Streaming TTS can provide the speech-generation layer of a Japanese conversational AI workflow.
Radiance provides 11 expression styles, including Conversational, Narrative, News, Happy, Sad, Enthusiastic, and others.
Yes. Radiance supports speed control from 0.25× to 4.0×.
Radiance supports PCM, WAV, MP3, OGG Opus, FLAC, μ-law, and A-law.
Yes. Radiance supports μ-law and A-law output formats used in telephony environments.
Japanese frequently contains Latin-script brand names, acronyms, and foreign terminology. Mixed-script inputs should be tested using your own production sentences to confirm pronunciation quality.
Yes. Japanese is also supported through Shunya's broader global speech-recognition coverage.
No. Parallax, Shunya's translation model, covers Indian languages and does not include Japanese. Radiance speaks Japanese text, but it does not translate it.
Yes. Radiance provides HTTP batch and WebSocket streaming APIs.
Get started
Don't judge Japanese TTS from our prepared sentence. Test yours. Try a customer response, Japanese name, product term,notification, narration, mixed Japanese-English sentence, voice-agent response, or text generated dynamicallyby your application. Hear how Shunya turns your Japanese text into speech.