| Model approach | One model (Seed Audio 1.0) renders speech, music, SFX, and ambient from a single prompt | Eight separate products (text-to-speech, speech-to-text, voice cloning, voice changer, story studio, audio separation, audio translation, sound effects) |
| Voice cloning | Zero-shot from a short clip you upload (up to 3 clips), referenced inline as @Audio1, no training step | Fast cloning from about 10-15 seconds of clean reference audio (also strong; a noisy clip needs re-recording) |
| Voice from an image | Define a voice from a character image, no recording needed | Not offered |
| Metering | 1 credit = 1 second; one-time credit packs that don't reset monthly | Roughly 600-625 credits per minute; monthly credits reset and unused credits are forfeited |
| Free start | Starter trial credits (about 30 seconds), cloning, preview, and download included | Free plan with 8,000 credits (about 7 minutes) |
| Languages | English and Chinese, with a wide range of accents and characters | Broader multilingual coverage across eight-plus languages |
| Voice library | Preset voices plus automatic voice selection | A 2,000,000-plus community voice library |
| Downloads | MP3, WAV, OGG, and PCM | Download and API export supported |
| 模型思路 | 單一模型(Seed Audio 1.0)用一個提示詞生成語音、音樂、音效和環境音 | 八個獨立產品(文字轉語音、語音轉文字、聲音複製、變聲器、故事工作室、音訊分離、音訊翻譯、音效) |
| 聲音複製 | 從你上傳的一小段音訊零樣本聲音複製(最多 3 段),在提示詞中以 @Audio1 引用,無需訓練步驟 | 用約 10-15 秒的乾淨參考音訊快速聲音複製(同樣出色;嘈雜的片段需要重新錄制) |
| 從圖片生成聲音 | 從角色圖片定義聲音,無需錄音 | 不提供 |
| 計費方式 | 1 個點數 = 1 秒;一次性點數方案,不按月清零 | 每分鐘約 600-625 個點數;月度點數會清零,未用完的點數作廢 |
| 免費起步 | 入門試用點數(約 30 秒),包含聲音複製、預覽和下載 | 免費方案含 8,000 個點數(約 7 分鐘) |
| 語言 | 英語和中文,涵蓋多種口音和角色 | 更廣的多語言覆蓋,支援八種以上語言 |
| 語音庫 | 預設聲音加自動聲音匹配 | 擁有超過 2,000,000 個社群聲音的語音庫 |
| 下載 | MP3、WAV、OGG 和 PCM | 支援下載和 API 匯出 |