RoleplayAdrianVampire archivist · The Vesper ArchivesYour Funeral Is TomorrowYour funeral is tomorrow. The invitation arrived today.0:00 / 0:00
Roleplay & Companions
Give the characters in your chat, story, or companion app a voice with BreezeBlue. Hear how different personalities and scenes shape the delivery, and explore what could fit your project.
Choose an example
Preview a voice, then step into the scene for the full conversation. From expressive roleplay to easy, everyday company, hear what could fit your product.
RoleplayAdrianVampire archivist · The Vesper ArchivesYour Funeral Is TomorrowYour funeral is tomorrow. The invitation arrived today.
RoleplayRinRuins scout · The RiftlandsAbout That Missing GoldYou made it back alive. The gold didn’t. Somehow, Rin’s the one who’s annoyed.0:00 / 0:00
CompanionMayaEveryday chat companion · After WorkWrong ChatYou texted your boss instead of your mom. Maya thinks it can’t be that bad.0:00 / 0:00
A voice for the character. A performance for the moment.
Design the voice, direct its expression, and keep conversations responsive with low-latency and multilingual speech.
Voice Design
Create your character’s voice
Describe how the character should sound. Audition it with a fitting line—no reference recording needed.
Casting & preset voices
- Recording takes time and budget.
- A preset may be close, but not quite your character.
BreezeBlue Voice Design
- Start with the role; add texture, accent and usual pace.
- Try a role-matched line. Refine one detail at a time.
- Save the voice you like and reuse it for new dialogue.
Different characters. Distinct voices.
Voice character
A mature male voice with a low register, slightly grainy texture, clear articulation and an unhurried pace.
(Speak quietly, as though this were quite ordinary. Keep it conversational, with a trace of dry humor.)
The book you wanted came back on its own last night. Don't open it yet. (chuckles) It hasn't decided whether it wants you reading it.
Voice Direction
Give every line a performance
Keep the chosen voice. Use an instruction to explain the moment and what the speaker wants to convey.
Leaving performance to the script
- Scripts can leave emotion unclear.
- Retakes mean briefing and recording again.
BreezeBlue Voice Direction
- Set the situation and intent. Pick two or three key cues.
- Specify intensity, pauses or emphasis—not just adjectives.
- Keep the voice and line fixed. Adjust direction and listen.
One line. Three situations.
The situation
The partner you feared you had lost has made it back safely.
(Begin with quiet disbelief, then let relief warm the ending, with a hint of reproach to hide your worry.)
[sigh] You actually came back.
Multilingual
Bring characters to more people
Localize dialogue for new audiences.
Separate language productions
- Casting adds work per market.
- Literal translation loses nuance.
BreezeBlue multilingual speech
- Create speech in 50+ languages with Breeze TTS 2 Multilingual.
- Shape delivery for each audience.
- Manage versions through one API.
One character. More languages.
The line
Take a seat. What’s the first thing you need to get off your chest today? I’m listening.
(Warm, casual, and genuinely curious.)
Take a seat. What's the first thing you need to get off your chest today? I'm listening.
Low Latency
Keep the character close to the conversation
Follow-up questions, hesitation and reassurance need a voice that keeps up. Stream speech so the character’s next reply can start sooner.
Generate, then play
- Full-clip generation keeps listeners waiting.
- Average latency hides slower replies.
BreezeBlue low-latency streaming
- Start returning audio with low time to first byte.
- Start playback as audio arrives.
- Keep questions, arguments and casual replies flowing.
Keep the character close to the conversation
Follow-up questions, hesitation and reassurance need a voice that keeps up. Stream speech so the character’s next reply can start sooner.
≤25msTime to first byte
Companion conversation · After work
- 1The openingA friend wants to talk about the hardest part of their day.
- 2The replyUse Maya’s familiar voice with an easy, attentive invitation to talk.
- 3The next turnReuse the realtime connection for the lines that follow.
Carry the same familiar voice from the first invitation into the conversation that follows.
Frequently asked questions
Character voices, AI companions and product integration
Which roleplay and AI companion experiences can use BreezeBlue?
BreezeBlue supports voice for character chat, interactive stories, game characters and everyday companions. The examples here show different personalities and situations. Bring your own character brief and dialogue to explore a similar experience for your product.
Can I design a custom AI character voice without a recording?
Yes. Voice Design generates previews from a written description without reference audio. Describe audible traits such as age, texture, accent and pacing. Audition a line, refine the description, and save the voice for new dialogue. Build the description around the qualities listeners can hear, then try it with dialogue from your character.
How can I keep a character’s voice recognizable across new dialogue?
Reuse the saved voice or voice ID for each new line, then adjust delivery instructions to fit the scene. Listen to the lines together to shape a consistent character across changing emotions and scenes.
Can instructions control emotion and delivery without changing the voice?
Yes. Choose a voice, then write a director’s note with the emotion, intent and pacing you want. Keep Instruction separate from the script and write it in the same language. For a local vocal action such as a sigh or chuckle, place a supported Audio tag at that point in the script. You do not need to redesign the voice for each performance.
How does BreezeBlue’s low latency help real-time character conversations?
Low time to first byte starts returning speech sooner, while streaming carries the rest of the dialogue. Begin playback as audio arrives to keep follow-up questions, playful exchanges and emotional replies connected. The same saved character voice and direction can carry through the conversation.
Can I create character voices in English, Chinese and other languages?
Breeze TTS 2 Multilingual supports 50+ languages. Available languages depend on the selected model. The multilingual panel on this page uses one Maya voice for Chinese, English, Japanese and Korean dialogue. Adapt the script and cultural details for each audience, then review pronunciation and naturalness.
How do I add character speech to my existing application?
Use the BreezeBlue API or SDK to generate speech with saved voices, dialogue and delivery instructions. The realtime interface accepts incremental text and streams audio back. Use a saved voice and scene-specific direction to bring new character dialogue to life.
Can I try my own dialogue instead of these examples?
Yes. Open Voice Design in BreezeBlue and design or choose a voice for your own lines. Start with a short exchange, compare the delivery, and save the voice that fits your character for future dialogue.
Find the voice for your characters
Bring your character and a few lines to BreezeBlue. Design a voice and shape its delivery—or talk to us about adding voice to your product.