🔥 Black Friday AI API deals are live· New drops every MondayEnds in--d --:--:--See deals →

Gemini 3.8 Flash TTS: AI 音频与音乐

使用 Gemini 3.8 Flash TTS 生成富有表现力的多说话人语音。具备录音棚级音质与长篇内容稳定性。免费试用,按使用量付费。 Gemini 3.8 Flash TTS 是 Google 新一代文本转语音模型,旨在提供录音棚级音质、富有表现力的演绎和长篇内容稳定性。可定义任意数量的说话者,为每位说话者设置独特的声音、口音、情绪风格和语速,再通过有序的对话轮次编写对话,并插入 [whispers] 或 [shouting] 等语气标签。它适用于角色对话、游戏和视频配音、播客及旁白,可快速呈现自然且富有表现力的表达。若高负载任务需要更低延迟和成本,请参阅 Gemini 3.8 Flash Lite TTS;若需要录音棚级高端输出,请参阅 Gemini 2.5 Pro TTS。

交互式模型控制

使用 Gemini 3.8 Flash TTS 生成富有表现力的多说话人语音。具备录音棚级音质与长篇内容稳定性。免费试用,按使用量付费。

📝

概览

关于此模型

Gemini 3.8 Flash TTS 是 Google 新一代文本转语音模型,旨在提供录音棚级音质、富有表现力的演绎和长篇内容稳定性。可定义任意数量的说话者,为每位说话者设置独特的声音、口音、情绪风格和语速,再通过有序的对话轮次编写对话,并插入 [whispers] 或 [shouting] 等语气标签。它适用于角色对话、游戏和视频配音、播客及旁白,可快速呈现自然且富有表现力的表达。若高负载任务需要更低延迟和成本,请参阅 Gemini 3.8 Flash Lite TTS;若需要录音棚级高端输出,请参阅 Gemini 2.5 Pro TTS。

1Games & Interactive: Voice NPCs and branching dialogue with distinct, emotionally-directed characters.
2Video & Shorts: Generate expressive, studio-grade voiceovers and character lines for content production.
3Podcasts: Produce multi-host conversational audio from a written script.
4Prototyping: Iterate quickly on tone, accent, and pacing before a final high-fidelity render.
💰

价格与价值

成本分析

muapiapp$0.015 per 1,000 characters

Pay per generation, no subscription. Matches Google's official Gemini 3.8 Flash TTS audio-output rate.

Fal.ai暂不可用

不提供此模型。

Replicate暂不可用

不提供此模型。

** 竞品价格根据相似模型架构和使用层级估算。

⚙️

技术详情

配置参数

说话人array

说话人声音配置列表。每一轮对话通过 ID 引用一个说话人。

默认值[object Object],[object Object]
对话轮次array

按顺序排列的对话行列表。每一轮的 speaker_id 必须匹配上面定义的说话人。文本可以包含 [shouting] 或 [whispers] 等语气标签。

默认值[object Object],[object Object],[object Object],[object Object]
场景string

可选的 scene description that sets the acoustic setting, e.g. "A quiet, warm room with a fireplace crackling softly."

默认值
示例上下文string

可选的 overall tone/style, e.g. "Audiobook style narration. Tone is gentle and inviting."

默认值
温度number

采样温度(0-2)。数值越高,表达方式越丰富。

默认值1
📖

实施指南

开发者文档

使用方法 Gemini 3.8 Flash TTS

  1. 定义说话者: 在其中为每种声音添加一项 speakers 列表。为每位说话者设置 speaker_id in "Speaker N" 格式,选择 voice_name, ,并设置 accent, style, and pace. 还可添加 audio_profile 用于描述人物设定。

  2. 编写对话: 在 dialogue_turns 中填写按顺序排列的台词。 每轮对话的 speaker_id 必须与已定义的说话者匹配。 可在文本中直接插入语气标签,例如 [whispers], [shouting], or [determination].

  3. 设置氛围(可选): 使用 scene 描述声音环境,并使用 sample_context 设置整体旁白基调。 调整 temperature (0-2) 以控制表达变化程度。

  4. 提交并轮询: POST 请求,然后轮询 /predictions/{request_id}/result 直到状态变为 completed 以获取音频 URL。

curl -X POST https://api.muapi.ai/api/v1/gemini-3-8-flash-tts \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "speakers": [
      {"speaker_id": "Speaker 1", "voice_name": "Fenrir", "accent": "British (RP)", "style": "Deadpan", "pace": "Natural"},
      {"speaker_id": "Speaker 2", "voice_name": "Puck", "accent": "American (Gen)", "style": "Empathetic", "pace": "Staccato"}
    ],
    "dialogue_turns": [
      {"speaker_id": "Speaker 1", "text": "[shouting] Halt, traveler!"},
      {"speaker_id": "Speaker 2", "text": "[determination] I carry a message for the elder."}
    ],
    "scene": "A dark, crumbling dungeon.",
    "temperature": 1
  }'
❓

常见问题

常见问答

How do I create a multi-speaker conversation?

Define each voice in the `speakers` list with a unique `speaker_id` (in "Speaker N" format), then reference those IDs in your `dialogue_turns`. 该model renders each line in that speaker's assigned voice, accent, style, and pace.

可以控制情绪和语速吗?

可以。 每个speaker has `style` (Vocal Smile, Newscaster, Whisper, Empathetic, Promo/Hype, Deadpan) and `pace` (Natural, Rapid Fire, 该Drift, Staccato) settings. 您可以also embed inline tone tags like [whispers], [shouting], or [determination] directly in the dialogue text.

支持哪些声音和口音?

There are 30 prebuilt voices (Zephyr, Fenrir, Puck, Kore, and more) and 8 accents including Neutral, American variants, British (RP), British (Brixton), Transatlantic, and Australian.

如何计算价格?

价格是based on the total number of characters across all dialogue turns, at $0.015 per 1,000 characters — matching Google's official Gemini 3.8 Flash TTS audio-output rate. You pay per generation with no subscription.

每行的最大长度是多少?

每个dialogue turn's text can be up to 10,000 characters. 您可以include multiple turns to build longer conversations.

How does this differ from Gemini 3.1 Flash TTS and Gemini 3.8 Flash Lite TTS?

Gemini 3.8 Flash TTS 是Google's newer generation Flash tier, tuned for studio-grade voice fidelity, expressive acting, and long-form stability at a lower cost than the 3.1 line. Gemini 3.8 Flash Lite TTS trades a little fidelity for even lower latency and cost on high-throughput, conversational workloads.