{"ok":true,"province":"tts","name":"音声合成（TTS）","name_en":"Text-to-Speech","level":1,"criteria":[{"id":"tts_natural","label":"自然さ","label_en":"Naturalness","direction":"higher","description":"","unit":null,"kind":"editorial"},{"id":"tts_voices","label":"声の種類・多言語","label_en":"Voices/langs","direction":"higher","description":"","unit":null,"kind":"editorial"},{"id":"tts_clone","label":"声クローン","label_en":"Voice cloning","direction":"higher","description":"","unit":null,"kind":"editorial"},{"id":"tts_speed","label":"速度（低遅延）","label_en":"Speed","direction":"higher","description":"","unit":null,"kind":"editorial"},{"id":"tts_price","label":"料金（相対）","label_en":"Price (rel)","direction":"lower","description":"従量課金APIの公称単価（$/100万文字・標準品質グレード基準）。サブスク型は単価比較が成立しないため欠測「—」とし、料金の方式を事実メモに記載。","unit":"$/100万文字","kind":"fact"},{"id":"tts_open","label":"オープン性","label_en":"Openness","direction":"higher","description":"","unit":null,"kind":"editorial"}],"items":[{"id":"t_eleven","name":"ElevenLabs","name_en":"ElevenLabs","provider":"ElevenLabs","jurisdiction":"US","note":"最自然・声クローンの定番","note_en":"Most natural; the go-to for voice cloning","url":"https://elevenlabs.io","shop_url":null,"metrics":{"tts_natural":95,"tts_voices":90,"tts_clone":92,"tts_speed":80,"tts_open":5},"facts":[{"label":"料金の方式","text":"サブスクリプション制（月額プランのクレジット消費）。文字単価換算では業界最高価格帯（概ね$100〜300/100万文字相当）だが品質・クローン機能が売り","source":"ElevenLabs公式プラン＋業界比較資料（2026-09-02確認）"}],"sources":{"tts_natural":"editorial: 最自然と記される通り多言語でも人間らしい抑揚に定評があり、候補内最上位のため高め（2026-06初評）","tts_voices":"editorial: 多言語対応と声ライブラリの広さで知られ、候補内ではAzureと並ぶ最上位のため高め（2026-06初評）","tts_clone":"editorial: 声クローンの定番と記される通りこの分野では代名詞的存在で、候補内最上位のため高め（2026-06初評）","tts_speed":"editorial: 低遅延モードも備えるが速度特化のCartesiaには及ばず、候補内では中位のため中位（2026-06初評）","tts_open":"editorial: モデル非公開の完全商用サービスでオープン性は皆無に近く、候補内最低水準のため低め（2026-06初評）","料金の方式":"ElevenLabs公式プラン＋業界比較資料（2026-09-02確認）"}},{"id":"t_openai","name":"OpenAI TTS","name_en":"OpenAI TTS","provider":"OpenAI","jurisdiction":"US","note":"手軽・統合","note_en":"Simple and well integrated","url":"https://openai.com","shop_url":null,"metrics":{"tts_natural":88,"tts_voices":70,"tts_clone":40,"tts_speed":82,"tts_open":5,"tts_price":15},"facts":[],"sources":{"tts_natural":"editorial: 手軽・統合と記される汎用APIながら自然さは高く、ElevenLabsに次ぐ2位のため高め（2026-06初評）","tts_voices":"editorial: 提供ボイスは少数のプリセットが中心で多言語の声の幅は狭く、候補内では下位のため低め（2026-06初評）","tts_clone":"editorial: 任意の声のクローン機能を公式には提供しない方針の業者で、候補内では下位のため低め（2026-06初評）","tts_speed":"editorial: 汎用APIとして応答は軽快で、候補内ではCartesiaとKokoroに次ぐ3位のためやや高め（2026-06初評）","tts_open":"editorial: モデル非公開のクラウドAPI提供のみでオープン性はなく、候補内最低水準のため低め（2026-06初評）","tts_price":"OpenAI公式 tts-1標準 $15/1M文字（HD版は$30・2026-09-02確認）"}},{"id":"t_google","name":"Google Cloud TTS","name_en":"Google Cloud TTS","provider":"Google","jurisdiction":"US","note":"多言語・安定","note_en":"Multilingual; stable","url":"https://cloud.google.com/text-to-speech","shop_url":null,"metrics":{"tts_natural":82,"tts_voices":88,"tts_clone":50,"tts_speed":80,"tts_open":5,"tts_price":16},"facts":[],"sources":{"tts_natural":"editorial: 安定した品質だが表現力特化の新興勢には一歩譲り、候補内では中位下の位置のため中位（2026-06初評）","tts_voices":"editorial: 多言語・安定と記される通り対応言語と声数の広さが強みで、候補内では3位のため高め（2026-06初評）","tts_clone":"editorial: カスタム音声の作成は法人向け限定の位置づけで手軽に使えず、候補内では中位のため中位（2026-06初評）","tts_speed":"editorial: 大手クラウドらしく応答は安定するが低遅延特化型には及ばず、候補内では中位のため中位（2026-06初評）","tts_open":"editorial: モデル非公開の商用クラウドAPI提供でオープン性はなく、候補内では最低水準のため低め（2026-06初評）","tts_price":"Google Cloud公式 Neural2 $16/1M文字（Standard/WaveNetは$4・Chirp3HDは$30・2026-09-02確認）"}},{"id":"t_azure","name":"Azure TTS","name_en":"Azure TTS","provider":"Microsoft","jurisdiction":"US","note":"多言語・企業向け","note_en":"Multilingual; enterprise-oriented","url":"https://azure.microsoft.com/products/ai-services/text-to-speech","shop_url":null,"metrics":{"tts_natural":83,"tts_voices":90,"tts_clone":70,"tts_speed":80,"tts_open":5,"tts_price":16},"facts":[],"sources":{"tts_natural":"editorial: 多言語・企業向けと記される安定品質だが表現力の上位には譲り、候補内中位のため中位（2026-06初評）","tts_voices":"editorial: 対応言語と声数の広さは業界屈指で、候補内ではElevenLabsと並ぶ最上位のため高め（2026-06初評）","tts_clone":"editorial: 企業向けカスタム音声を提供するが利用審査が前提で、候補内では中位上のためやや高め（2026-06初評）","tts_speed":"editorial: 大手クラウドとして応答は安定するが低遅延特化型には及ばず、候補内では中位のため中位（2026-06初評）","tts_open":"editorial: モデル非公開の商用クラウドサービスでオープン性はなく、候補内では最低水準のため低め（2026-06初評）","tts_price":"Azure AI Speech公式 Neural音声の標準従量単価（2026-09-02確認）"}},{"id":"t_cartesia","name":"Cartesia","name_en":"Cartesia","provider":"Cartesia","jurisdiction":"US","note":"超低遅延","note_en":"Ultra-low latency","url":"https://cartesia.ai","shop_url":null,"metrics":{"tts_natural":86,"tts_voices":72,"tts_clone":75,"tts_speed":95,"tts_open":10},"facts":[{"label":"料金の方式","text":"サブスクリプション＋クレジット制。低遅延（リアルタイム向け）が特長で単価比較には不向き","source":"Cartesia公式（2026-09-02確認）"}],"sources":{"tts_natural":"editorial: 超低遅延と記される速度特化ながら自然さも高い水準で、候補内では3位のためやや高め（2026-06初評）","tts_voices":"editorial: 声数と対応言語の幅は大手クラウドの網羅性には及ばず、候補内では中位下の位置のため中位（2026-06初評）","tts_clone":"editorial: 即時クローンに対応するが定番のElevenLabsらには一歩譲り、候補内3位のためやや高め（2026-06初評）","tts_speed":"editorial: 超低遅延と記される通りリアルタイム会話用途に特化した型で、候補内最上位のため高め（2026-06初評）","tts_open":"editorial: 商用API中心だが完全非公開の各社よりわずかに開かれ、候補内2位ながら水準は低め（2026-06初評）","料金の方式":"Cartesia公式（2026-09-02確認）"}},{"id":"t_playht","name":"PlayHT","name_en":"PlayHT","provider":"PlayHT","jurisdiction":"US","note":"クローンと声数","note_en":"Voice cloning and a large number of voices","url":"https://play.ht","shop_url":null,"metrics":{"tts_natural":84,"tts_voices":78,"tts_clone":82,"tts_speed":78,"tts_open":5},"facts":[{"label":"料金の方式","text":"サブスクリプション制（プランごとの文字数上限）","source":"PlayHT公式（2026-09-02確認）"}],"sources":{"tts_natural":"editorial: 自然さは水準以上だが最上位勢の表現力には一歩届かず、候補内では中位の位置のため中位（2026-06初評）","tts_voices":"editorial: 声数の多さを売りとするが大手クラウドの多言語網には及ばず、候補内では中位のため中位（2026-06初評）","tts_clone":"editorial: クローンと声数と記される通り声のクローンが主力機能の業者で、候補内2位のため高め（2026-06初評）","tts_speed":"editorial: 応答は実用水準だが低遅延特化型や大手勢には一歩譲り、候補内では下位のためやや低め（2026-06初評）","tts_open":"editorial: モデル非公開の商用クラウドサービスであり、オープン性はなく候補内最低水準のため低め（2026-06初評）","料金の方式":"PlayHT公式（2026-09-02確認）"}},{"id":"t_murf","name":"Murf","name_en":"Murf","provider":"Murf","jurisdiction":"US","note":"ナレーション制作向け","note_en":"For narration production","url":"https://murf.ai","shop_url":null,"metrics":{"tts_natural":80,"tts_voices":75,"tts_clone":40,"tts_speed":75,"tts_open":5},"facts":[{"label":"料金の方式","text":"サブスクリプション制（編集スタジオ込みのSaaS型）","source":"Murf公式（2026-09-02確認）"}],"sources":{"tts_natural":"editorial: ナレーション向けの聞き取りやすさを重視し表現力上位には譲り、候補内下位のため低め（2026-06初評）","tts_voices":"editorial: ナレーター系の声を幅広く揃えるが多言語の網羅性は大手に及ばず、候補内中位のため中位（2026-06初評）","tts_clone":"editorial: 編集画面でのナレーション制作が主でクローンは主機能ではなく、候補内下位のため低め（2026-06初評）","tts_speed":"editorial: 編集画面型の制作スタジオでリアルタイム応答は主眼でなく、候補内では最下位のため低め（2026-06初評）","tts_open":"editorial: モデル非公開の商用制作サービスでありオープン性はなく、候補内では最低水準のため低め（2026-06初評）","料金の方式":"Murf公式（2026-09-02確認）"}},{"id":"t_kokoro","name":"Kokoro TTS","name_en":"Kokoro TTS","provider":"(open)","jurisdiction":"US","note":"軽量・オープンで安い","note_en":"Lightweight, open, and cheap","url":"https://huggingface.co/hexgrad/Kokoro-82M","shop_url":null,"metrics":{"tts_natural":78,"tts_voices":55,"tts_clone":30,"tts_speed":85,"tts_open":95,"tts_price":0},"facts":[],"sources":{"tts_natural":"editorial: 軽量・オープン系としては健闘するが商用勢の表現力には及ばず、候補内最下位のため低め（2026-06初評）","tts_voices":"editorial: 軽量設計ゆえ提供ボイス数と対応言語の幅が限られ、候補内で最も少ない水準のため低め（2026-06初評）","tts_clone":"editorial: 任意の声を再現するクローン機能を持たない軽量モデルで、候補内では最下位のため低め（2026-06初評）","tts_speed":"editorial: 軽量でローカルでも高速に動く型として知られ、候補内Cartesiaに次ぐ2位のため高め（2026-06初評）","tts_open":"editorial: オープンと記される通り重みが公開されローカル実行も自由な候補内唯一の公開系のため高め（2026-06初評）","tts_price":"オープンソース（Apache系）＝API課金なし。自前GPU/CPUの計算コストは別途かかる（2026-09-02確認）"}}],"profiles":[{"id":"tts_nat","name":"自然さ重視","name_en":"Naturalness","weights":{"tts_natural":0.45,"tts_voices":0.15,"tts_clone":0.15,"tts_speed":0.1,"tts_price":0.1,"tts_open":0.05}},{"id":"tts_multi","name":"多言語・声数","name_en":"Multilingual","weights":{"tts_natural":0.25000000000000006,"tts_voices":0.45000000000000007,"tts_clone":0.10000000000000002,"tts_speed":0.10000000000000002,"tts_price":0.10000000000000002}},{"id":"tts_clone","name":"声クローン","name_en":"Cloning","weights":{"tts_natural":0.25,"tts_voices":0.1,"tts_clone":0.45,"tts_speed":0.1,"tts_price":0.1}},{"id":"tts_low","name":"低遅延","name_en":"Low-latency","weights":{"tts_natural":0.25,"tts_voices":0.1,"tts_clone":0.05,"tts_speed":0.45,"tts_price":0.15}},{"id":"tts_open","name":"自前運用","name_en":"Self-host","weights":{"tts_natural":0.2,"tts_speed":0.1,"tts_price":0.25,"tts_open":0.45}},{"id":"tts_bal","name":"バランス","name_en":"Balanced","weights":{"tts_natural":0.18,"tts_voices":0.17,"tts_clone":0.15,"tts_speed":0.17,"tts_price":0.18,"tts_open":0.15}}],"articles":[{"slug":"choose-tts-2026","title":"音声合成（TTS）の選び方：用途で決まる5つの軸","dek":"声のAIはもう一つじゃない。ナレ/多言語/感情、用途で最適は変わる。"}],"terms":[],"campaigns":[]}