{"schema_version":"1.0","checked_at":"2026-09-14","publisher":"CastReader","evidence_type":"first-party compilation of official published documents","independent":false,"competitor_audio_tested":false,"quality_ranking":null,"source_policy":"Each row links to primary documentation. Prices and limits are dated, not a live quote. Unknown does not mean unsupported.","calculation_method":"Use the same CRLF-to-LF, NFC-normalized, trimmed text. CastReader counts code points; Fish counts UTF-8 bytes. Character-priced competitor estimates assume one billed character per code point: confirm exact provider metering. Do not convert credits or output audio tokens from text alone.","exclusions":["Free credits and free models","Taxes, discounts, minimum top-ups and unused balances","Subscription/overage terms","Cloning preparation or ancillary API charges","Application hosting, storage, bandwidth, developer time and regenerations"],"note":"Fish lists s2.1-pro-free at $0; this paid-rate calculator is not proof that CastReader is cheapest. Price comparisons do not establish equal quality, latency or features.","providers":[{"id":"castreader","name":"CastReader","model":"clone-v1","price":["$8 / 1M normalized characters","$8 / 百万规范化字符"],"meter":"codepoints","ratePerMillion":8,"admission":["Verified identity → self-service activation; one 3,000-character trial with expiry/daily limits. No monthly API subscription.","验证身份后自助开通；一次 3,000 字符试用，受有效期与每日限制。无 API 月费。"],"fit":["Short, prepared multilingual audio and authorized voice cloning. Saved MP3/WAV; recoverable queued jobs.","短篇预生成多语言音频与授权声音克隆。保存 MP3/WAV，可恢复的排队任务。"],"boundary":["10 configured languages; 500 characters/queued job, workspace concurrency 1. No streaming, realtime conversation, emotion parameter or latency SLA.","配置 10 种语言；排队任务最多 500 字符，工作区并发 1。无流式、实时对话、情绪参数或延迟 SLA。"],"source":"https://voice.castreader.com/integration-manifest.json","docs":"https://voice.castreader.com/docs/overview/capabilities"},{"id":"elevenlabs","name":"ElevenLabs","model":"Flash / Turbo","price":["$50 / 1M characters","$50 / 百万字符"],"meter":"characters","ratePerMillion":50,"admission":["Pay-as-you-go is available; check free/startup allowances and account terms separately.","已有按量付费；另行核对免费、初创优惠与账户条款。"],"fit":["A candidate when faster delivery, more languages or longer individual inputs are required.","需要更快交付、更多语言或更长单次输入时的候选。"],"boundary":["Official Flash/Turbo listing: 32 languages, 40,000-character input limit. v2/v3 have different prices and capabilities; not interchangeable benchmarks.","官方 Flash/Turbo 列出 32 种语言、40,000 字符输入上限。v2/v3 价格与能力不同，不应混为同一模型评测。"],"source":"https://elevenlabs.io/pricing/api","docs":"https://elevenlabs.io/text-to-speech-api"},{"id":"fish","name":"Fish Audio","model":"s2.1-pro (paid)","price":["$15 / 1M UTF-8 bytes","$15 / 百万 UTF-8 字节"],"meter":"utf8_bytes","ratePerMillion":15,"admission":["Pay-as-you-go. The official price list also includes s2.1-pro-free at $0; verify free-model access and limits before choosing paid usage.","按量付费。官方还列出 $0 的 s2.1-pro-free；选择付费前先核对免费模型的准入与限制。"],"fit":["Official Python/TypeScript clients, cloning, REST and WebSocket streaming.","官方 Python/TypeScript 客户端，声音克隆、REST 与 WebSocket 流式。"],"boundary":["Bytes are not characters. The pricing table lists starter concurrency 5; account/model-specific free limits and production terms require confirmation.","字节不等于字符。价格表列出入门并发 5；免费模型具体限制与生产条款需另行确认。"],"source":"https://docs.fish.audio/developer-guide/models-pricing/pricing-and-rate-limits","docs":"https://fish.audio/developers/"},{"id":"deepgram","name":"Deepgram","model":"Aura-2","price":["$30 / 1M characters (PAYG)","$30 / 百万字符（按量付费）"],"meter":"characters","ratePerMillion":30,"admission":["PAYG and Growth tiers have different rates. Free credits and other models are separate.","按量付费与 Growth 档单价不同；免费额度和其他模型另计。"],"fit":["Consider alongside other speech-service options; evaluate the chosen voice and language in your own workload.","可纳入语音服务选型；使用自己的工作负载评估具体音色与语言。"],"boundary":["Aura-1 is also listed at $15/1M characters. Flux has a separate price/promotion; this row does not mix these models or establish clone parity.","Aura-1 另列 $15/百万字符；Flux 使用独立价格与优惠。本行不混用模型，也不证明克隆能力等价。"],"source":"https://deepgram.com/pricing"},{"id":"cartesia","name":"Cartesia","model":"Pro plan","price":["$5/month · 100K credits","$5/月 · 10 万 credits"],"meter":"credits","admission":["A Free plan exists. Pro includes instant voice cloning and commercial use in the published plan.","提供 Free 档；公开 Pro 方案包含即时克隆与商业使用。"],"fit":["Evaluate when its voice/cloning workflow matches the application; compare plan utilization, not only the headline fee.","音色或克隆流程适配时可评估；比较套餐利用率，不只比较标价。"],"boundary":["No character-equivalent estimate here: confirm model credit conversion, overage, term and account limits first.","这里不强行换算字符价格：先确认模型 credits 换算、超额、订阅周期与账户限制。"],"source":"https://www.cartesia.ai/pricing"},{"id":"openai","name":"OpenAI","model":"gpt-4o-mini-tts","price":["$0.60/1M text input tokens + $12/1M audio output tokens","$0.60/百万输入文本 tokens + $12/百万输出音频 tokens"],"meter":"tokens","admission":["API access and rate limits depend on the account. Custom voices have a separate eligibility and consent process.","API 准入与限流依账户而定；自定义声音有单独的准入和授权流程。"],"fit":["Streaming speech and delivery instructions; a candidate when an existing application already uses OpenAI.","流式语音与表达指令；已有 OpenAI 接入的应用可以纳入候选。"],"boundary":["2,000 input tokens maximum on the model page. Output audio tokens are not known from character count alone; no fixed $/1M-character conversion.","模型页标注最多 2,000 输入 tokens。仅凭字符数无法知道输出音频 tokens，不给出固定的每百万字符换算。"],"source":"https://developers.openai.com/api/docs/models/gpt-4o-mini-tts","docs":"https://developers.openai.com/api/docs/guides/text-to-speech"}]}