Live Streaming

Give virtual hosts and stream co-hosts a recognizable voice with BreezeBlue. Bring character to chat replies, segment introductions and audience games. Direct each line's emotion, keep pace with streaming speech, and connect with viewers from different regions in their own language.

Try Voice Design Contact

The host steps away. The conversation keeps going.

A short water break. One AI co-host. Hear Pip join chat's jokes, welcome the host back, and try to move on.

Host note

“Grabbing some water. Pip, keep chat company.”

Scripted scene demo

● Host away

Taking overA pleased chuckle, then a conspiratorial whisper to chat.

Host awayPip is reading chat

Pip replies

All right, I'm in charge for thirty seconds. What have you always wanted to ask, but never with the streamer here?

Chat

The host is getting water. Pip has chat to himself.

0:00 / 1:12

Scene moments

A voice that belongs in your stream

Shape the voice, direct each reply, and let your co-host speak to more viewers.

Voice Design

A voice viewers recognize

A co-host's voice is how viewers first meet the character. Compare Pip's lively sound with a gentle female voice and a playful male voice, all answering the same chat prompt. Find a personality that fits your stream.

The challenge

A generic announcement voice gives viewers little to recognize.
An overpowering voice competes with the host's style.
Changing voices with every segment makes the character unfamiliar.

With BreezeBlue

Describe texture and pace to create a distinct personality.
Compare three voice presets on the same chat reply.
Save the co-host's voice for a familiar sound each stream.

Meet your co-host's voice

The voice

Pip is the familiar, playful co-host: bright, nimble and a little pleased with himself. Let chat’s question spark his interest, take a moment to think, then tease familiar viewers for offering the answer too soon.

Now that's a good question. Let me think. Don't answer it for me, chat.

0:00 / 0:06

Try Voice Design Read Docs

Voice Direction

The same words, a different moment

Chat might pick Pip's favourite minigame, ask him to sing the opening tune, or choose a challenge he struggles with. The same reply becomes delighted, self-conscious or playfully defiant as the situation changes.

The challenge

Surprise, hesitation and defiance sound like the same announcement.
The words fit, but the reply feels detached from the moment.
Every new reaction requires switching the character's voice.

With BreezeBlue

Direct a reply through the situation and the character's intent.
Express changing emotions through emphasis, pace and pauses.
Keep Pip's voice across all three reactions.

One line. Three reactions.

The moment

Chat picks the minigame Pip secretly hoped to play. The question opens with delighted surprise, as if viewers have read his mind. Let the reply accept eagerly, with a smile in the voice and a clean, confident finish.

You're really picking that? All right, let's do it.

0:00 / 0:03

Try Voice Direction Read Docs

Multilingual

One co-host, more audiences

Invite viewers to suggest what they would like to see next time. Hear the same closing question in Chinese, English, Japanese and Korean, keeping Pip's curiosity and playful tone across audiences.

The challenge

Literal translations turn casual banter into stiff announcements.
Changes in formality can pull the character out of voice.
Viewers miss an invitation they cannot understand.

With BreezeBlue

Write natural local versions of the same stream moment.
Invite audiences in different regions with multilingual speech.
Shape each language's delivery around the same character.

One co-host. Four languages.

The line

The stream is winding down and Pip leaves one last question for chat. Invite ideas for the next stream with curiosity and a playful, conversational tone. Keep the question clear and leave room for viewers to answer.

One last question for chat: what would you like to see on the next stream?

0:00 / 0:04

View Multilingual Docs

Streaming speech

Let chat replies move with the stream

When the host steps away, short replies can keep the exchange going. Send text sentence by sentence and receive speech progressively, with a familiar voice across turns.

The challenge

Waiting for a complete recording misses the topic chat just raised.
Reopening a connection for every reply adds avoidable setup time.
The previous generation can continue after the host returns.

With BreezeBlue

Send text progressively and receive speech as it is generated.
Reuse the connection across nearby turns with the same co-host voice.
Update delivery between turns and cancel the active generation when needed.

Let chat replies move with the stream

When the host steps away, short replies can keep the exchange going. Send text sentence by sentence and receive speech progressively, with a familiar voice across turns.

The host gets water · Pip takes over

1Answer chatKeep Pip's bright, playful voice as he joins viewers teasing the host's “five minutes.”
2Keep talkingSend the next sentence when it is ready. Receive speech as it is generated, without preparing a complete audio file first.
3Hand backCancel the active generation turn when the host is ready to take over the conversation.

Reuse a realtime connection across nearby turns and update delivery instructions between them for banter, welcomes and transitions.

Get API Key Read Docs

Frequently asked questions

What speaking roles can an AI co-host bring to a stream?

A co-host can give chat replies, minigame introductions, welcomes and closing questions a distinct character voice. Start with the relationship between host and co-host, then choose a sound and pace that fit. A playful character like Pip brings personality through short, expressive replies.

How do viewers get to know one consistent co-host?

Define the voice's texture, age impression and rhythm in Voice Design, then save it for future content. Chat replies, segment introductions and language versions can all build around that voice. Direction instructions adjust the attitude of each line while preserving the character's familiar sound.

Can one reply sound delighted, shy or ready for a challenge?

Yes. Keep the chosen voice and line, and describe what just happened and how the co-host reacts. Pip's response to chat choosing a minigame demonstrates three attitudes through emphasis, pauses and pacing, making it easy to compare performances for the mood of your stream.

How do I create co-host speech for audiences in other languages?

Adapt wording and conversational habits for each audience, then generate the localized versions with multilingual speech. Keep short questions casual and inviting. The Chinese, English, Japanese and Korean examples here show Pip asking viewers what they want to see on the next stream.

How can streaming speech fit into a live setup?

Use streaming synthesis for a complete introduction, or a realtime connection for successive chat replies. Send text progressively and receive audio as it is generated. Reuse the connection across nearby turns and update delivery instructions between them. Cancel the active generation when the host needs to take the conversation back.

Give your stream a voice of its own

Start with a welcome, a chat reply or a segment introduction. Design a co-host voice your viewers will remember.

Try Voice Design Talk to us