LimitHit
语言: 中文

本页为机器翻译。如有错误，请发邮件至 [email protected]。

对比 · 15 家厂商 · 核对于

语音 AI API 价格

文字转语音按每 100 万字符计，语音转文字按每小时音频计。均为各厂商官网的按量付费标价。

最便宜的 TTS：每 100 万字符 $4（Google Cloud Standard, Amazon (AWS) Polly Standard）。最便宜的 STT：每小时 $0.04（Groq whisper-large-v3-turbo）。

文字转语音，每 100 万字符

每家厂商最便宜的现行模型 · 按量付费

  1. Google CloudStandard

    $4 每 100 万字符

  2. Amazon (AWS)Polly Standard

    $4 每 100 万字符

  3. SpeechmaticsSpeechmatics TTS

    $11 每 100 万字符

  4. OpenAItts-1 2027-01-06 停用

    $15 每 100 万字符

  5. Azure AI SpeechNeural

    $15 每 100 万字符

  6. DeepgramAura-1

    $15 每 100 万字符

  7. InworldTTS-2 Flash

    $15 每 100 万字符

  8. RimeMist v3

    $30 每 100 万字符

  9. ElevenLabsEleven v4 Turbo

    $40 每 100 万字符

  10. CartesiaSonic-3.6

    $65 每 100 万字符

语音转文字，每小时音频

每家厂商最便宜的现行模型 · 批量或实时

  1. Groqwhisper-large-v3-turbo

    $0.04 每小时音频

  2. SonioxSoniox async

    $0.10 每小时音频

  3. RimeRime STT · 实时

    $0.12 每小时音频

  4. SpeechmaticsMelia 1

    $0.12 每小时音频

  5. AssemblyAIUniversal-2

    $0.15 每小时音频

  6. InworldSTT 1 · 实时

    $0.15 每小时音频

  7. GladiaSolaria-3 · €0.16

    ≈$0.179 每小时音频

  8. Google CloudV2 dynamic batch

    $0.18 每小时音频

  9. Azure AI SpeechBatch transcription

    $0.18 每小时音频

  10. Rev.aiReverb

    $0.20 每小时音频

  11. ElevenLabsScribe v2

    $0.22 每小时音频

  12. DeepgramNova-3 Monolingual

    $0.258 每小时音频

  13. OpenAIgpt-transcribe

    $0.27 每小时音频

  14. Amazon (AWS)Transcribe batch

    $0.36 每小时音频

  15. CartesiaInk-2 · 实时

    $0.70 每小时音频

用量和套餐折扣见备注。按 token 计价的模型（Gemini TTS、gpt-4o-mini-tts）没有按字符的价格，未列入图表。€ 按 2026-10-22 欧洲央行汇率 1.1205 换算。

即将停用

全部文字转语音模型

点击列标题排序 · 各厂商详情见下方

文字转语音 API 价格：厂商、模型及停用日期、每 100 万字符价格和备注。列标题是排序按钮；按 token 计价的模型始终排在最后。
备注
Google CloudStandard $4$4 / 1M characters4M characters free a month.
Google CloudWaveNet $4$4 / 1M characters4M characters free a month.
Amazon (AWS)Polly Standard $4$4 / 1M characters5M characters free a month.
SpeechmaticsSpeechmatics TTS $11$0.011 per 1,000 charactersFirst 1M characters free.
OpenAItts-1 2027-01-06 停用$15$15.00 / 1M charactersReplacement: gpt-realtime-2.1-mini.
Azure AI SpeechNeural $15$15 / 1M charactersEast US. 0.5M characters free a month (F0).
DeepgramAura-1 $15$0.0150 / 1K characters
InworldTTS-2 Flash $15$15 / 1M charactersOn-Demand rate; $7 on Growth.
Google CloudNeural2 $16$16 / 1M characters1M characters free a month.
Amazon (AWS)Polly Neural $16$16 / 1M characters1M free a month for 12 months.
Azure AI SpeechNeural HD $22$22 / 1M charactersEast US.
InworldTTS-2 $25$25 / 1M charactersOn-Demand rate; $12.50 on Growth.
OpenAItts-1-hd 2027-01-06 停用$30$30.00 / 1M charactersReplacement: gpt-realtime-2.1-mini.
Google CloudChirp 3 HD $30$30 / 1M characters1M characters free a month.
Amazon (AWS)Polly Generative $30$30 / 1M characters100K free a month for 12 months.
DeepgramAura-2 $30$0.030 / 1K characters
RimeMist v3 $30$0.03 per 1K characters
ElevenLabsEleven v4 Turbo $40$0.04 / 1K characters
ElevenLabsFlash / Turbo v2.5 $40$0.04 / 1K characters
DeepgramFlux TTS $45$0.0450 / 1K characters$1 credit per $1 spent, up to $500, until Dec 31, 2026.
RimeCoda v2 $50$0.05 per 1K characters
CartesiaSonic-3.6 $65$65 / 1M credits (Pro overage)1 credit = 1 character. $38 per 1M on Scale.
ElevenLabsEleven v4 $80$0.08 / 1K characters
ElevenLabsEleven v3 $80$0.08 / 1K characters5,000 characters per request.
ElevenLabsMultilingual v2 $80$0.08 / 1K characters
Amazon (AWS)Polly Long-Form $100$100 / 1M characters500K free a month for 12 months.
Google CloudStudio $160$160 / 1M characters1M characters free a month.
OpenAIgpt-4o-mini-tts 2027-01-06 停用$0.60 / 1M text in · $12 / 1M audio out (tokens)Token-priced; no per-character rate. Snapshots 2025-03-20 and 2025-12-15 shut down Jan 6, 2027; replacement gpt-realtime-2.1-mini.
Google CloudGemini 3.8 Flash TTS 预览版$0.50 / 1M text in · $9 / 1M audio out (tokens)Rises to $1 / $18 on Jan 1, 2027. 25 audio tokens = 1 second, so ~$0.81 per hour of output audio plus input.
Google CloudGemini 2.5 Flash TTS $0.50 / 1M text in · $10 / 1M audio out (tokens)Token-priced; ~$0.90 per hour of output audio plus input.
SonioxSoniox TTS $4 / 1M text in · $21.50 / 1M audio out (tokens)Vendor estimate: about $0.70 per hour of generated speech.

全部语音转文字模型

点击列标题排序 · 各厂商详情见下方

语音转文字 API 价格：厂商、模型及停用日期、批量或实时、每小时音频价格和备注。列标题是排序按钮；没有每小时价格的模型始终排在最后。
备注
Groqwhisper-large-v3-turbo 批量$0.04$0.04 / hour
SonioxSoniox async 批量$0.10$0.10 / hour
Groqwhisper-large-v3 批量$0.111$0.111 / hour
RimeRime STT 实时$0.12$0.002 / min from Nov 20, 2026Free until Nov 19, 2026.
SonioxSoniox real-time 实时$0.12$0.12 / hour
SpeechmaticsMelia 1 批量$0.12$0.12 / hour
AssemblyAIUniversal-2 批量$0.15$0.15 / hour
AssemblyAIUniversal-Streaming 实时$0.15$0.15 / hour
InworldSTT 1 实时$0.15$0.15 / hourOn-Demand rate; $0.10/h on paid plans.
GladiaSolaria-3 批量≈$0.1790.16€ / hour (0–100 h)
OpenAIgpt-4o-mini-transcribe 2027-02-26 停用批量$0.18$0.003 / min (estimated)Replacement: gpt-transcribe or gpt-live-transcribe.
Google CloudV2 dynamic batch 批量$0.18$0.003 / min
Azure AI SpeechBatch transcription 批量$0.18$0.18 / hourEast US.
Rev.aiReverb 批量$0.20$0.20 / hourEnglish.
AssemblyAIUniversal-3.5 Pro 批量$0.21$0.21 / hour
ElevenLabsScribe v2 批量$0.22$0.22 / hourKeyterm prompting +$0.05/h.
SpeechmaticsStandard 批量$0.23$0.23 / hour
DeepgramNova-3 Monolingual 批量$0.258$0.0043 / min
OpenAIgpt-transcribe 批量$0.27$0.0045 / min (estimated)
DeepgramWhisper Large 批量$0.288$0.0048 / min
DeepgramNova-3 Monolingual (streaming) 实时$0.288$0.0048 / min (promo)Regular price $0.0077/min ($0.46/h).
Rev.aiReverb Foreign Language 批量$0.30$0.30 / hour
Rev.aiWhisper Large 批量$0.30$0.005 / min
GladiaSolaria-1 批量≈$0.3030.27€ / hour (0–100 h)
DeepgramNova-3 Multilingual 批量$0.312$0.0052 / min
OpenAIgpt-4o-transcribe 2027-02-26 停用批量$0.36$0.006 / min (estimated)Replacement: gpt-transcribe or gpt-live-transcribe.
OpenAIwhisper-1 2027-02-26 停用批量$0.36$0.006 / minReplacement: gpt-transcribe or gpt-live-transcribe.
Amazon (AWS)Transcribe batch 批量$0.36$0.0001 / second (us-east-1)$0.006/min. 60 min free a month for 12 months.
Azure AI SpeechFast transcription 批量$0.36$0.36 / hourEast US.
SpeechmaticsEnhanced 批量$0.38$0.38 / hour
ElevenLabsScribe v2 Realtime 实时$0.39$0.39 / hour
DeepgramFlux English 实时$0.39$0.0065 / min (promo)Regular price $0.0077/min ($0.46/h).
AssemblyAIUniversal-3.6 Pro Realtime 实时$0.45$0.45 / hour
Azure AI SpeechMAI-Transcribe-2 批量$0.54$0.54 / hourEast US, Foundry Models meter.
Amazon (AWS)Transcribe streaming 实时$0.60$0.0001667 / second (us-east-1)$0.01/min.
CartesiaInk-2 实时$0.703 credits / secondDerived: 10,800 credits per hour at the $65 per 1M Pro overage rate. Vendor says $0.39/h on Scale.
GladiaReal-time 实时≈$0.7620.68€ / hour
Google CloudV2 standard (incl. Chirp) 实时$0.96$0.016 / min (first 500K min)Drops to $0.004/min above 2M min.
Azure AI SpeechReal-time transcription 实时$1.00$1.00 / hourEast US. 5 h free a month (F0).
OpenAIgpt-live-transcribe 实时$1.02$0.017 / min (estimated)

各厂商详情

Groq

语音转文字
  • whisper-large-v3-turbo：$0.04 · 批量
  • whisper-large-v3：$0.111 · 批量
每小时音频
免费额度
未知 free plan exists with rate limits; no hour quota stated
备注
Hosted Whisper models. Per-hour cost from the speech-to-text docs.
  1. 访问 Groq ↗
  2. Speech-to-text docs ↗
  3. 核对于

Soniox

文字转语音
  • Soniox TTS：$4 / 1M text in · $21.50 / 1M audio out (tokens)
每 100 万字符
语音转文字
  • Soniox async：$0.10 · 批量
  • Soniox real-time：$0.12 · 实时
每小时音频
免费额度
未知 no free quota stated on the pricing page
备注
Diarization, language ID and translation included in the hourly rate. TTS is token-priced ($4 per 1M input text tokens, $21.50 per 1M output audio tokens), about $0.70 per hour of generated speech.
  1. 访问 Soniox ↗
  2. Pricing ↗
  3. 核对于

Rime

文字转语音
  • Mist v3：$30
  • Coda v2：$50
每 100 万字符
语音转文字
  • Rime STT：$0.12 · 实时
每小时音频
免费额度
$50 free credits. STT free until Nov 19, 2026.
备注
TTS metered by characters (~1K per minute).
  1. 访问 Rime ↗
  2. Pricing ↗
  3. 核对于

Speechmatics

文字转语音
  • Speechmatics TTS：$11
每 100 万字符
语音转文字
  • Melia 1：$0.12 · 批量
  • Standard：$0.23 · 批量
  • Enhanced：$0.38 · 批量
每小时音频
免费额度
$100 in credits. TTS: first 1M characters free.
备注
Pre-recorded rates shown; real-time rates are not in the page text. Volume discount up to 25%.
  1. 访问 Speechmatics ↗
  2. Pricing ↗
  3. 核对于

AssemblyAI

语音转文字
  • Universal-2：$0.15 · 批量
  • Universal-3.5 Pro：$0.21 · 批量
  • Universal-Streaming：$0.15 · 实时
  • Universal-3.6 Pro Realtime：$0.45 · 实时
每小时音频
免费额度
Free tier: up to 185 hours pre-recorded and 333 hours streaming, no card required.
备注
Multichannel audio is billed per channel.
  1. 访问 AssemblyAI ↗
  2. Pricing ↗
  3. 核对于

Inworld

文字转语音
  • TTS-2 Flash：$15
  • TTS-2：$25
每 100 万字符
语音转文字
  • STT 1：$0.15 · 实时
每小时音频
免费额度
On-Demand includes up to 70 min of TTS or 400 min of STT free.
备注
Paid plans cut rates: TTS-2 down to $12.50 and TTS-2 Flash to $7 per 1M characters, STT to $0.10/h. Enterprise from $5 per 1M.
  1. 访问 Inworld ↗
  2. Pricing ↗
  3. 核对于

Gladia

语音转文字
  • Solaria-3：≈$0.179 · 批量
  • Solaria-1：≈$0.303 · 批量
  • Real-time：≈$0.762 · 实时
每小时音频
免费额度
50€ free credits.
备注
Prices in € (≈$ at the ECB rate for ranking). Volume tiers drop Solaria-3 async to 0.12€/h.
  1. 访问 Gladia ↗
  2. Pricing ↗
  3. 核对于

Google Cloud

文字转语音
  • Standard：$4
  • WaveNet：$4
  • Neural2：$16
  • Chirp 3 HD：$30
  • Studio：$160
  • Gemini 3.8 Flash TTS：$0.50 / 1M text in · $9 / 1M audio out (tokens) 预览版
  • Gemini 2.5 Flash TTS：$0.50 / 1M text in · $10 / 1M audio out (tokens)
每 100 万字符
语音转文字
  • V2 dynamic batch：$0.18 · 批量
  • V2 standard (incl. Chirp)：$0.96 · 实时
每小时音频
免费额度
TTS: 4M Standard/WaveNet or 1M Chirp 3 HD/Neural2/Studio characters free a month. STT V1: 60 min free a month. Gemini TTS: no free tier.
备注
Gemini TTS is priced per token (25 audio tokens = 1 second). STT billed per second.
  1. 访问 Google Cloud ↗
  2. TTS pricing ↗
  3. STT pricing ↗
  4. 核对于

Azure AI Speech

文字转语音
  • Neural：$15
  • Neural HD：$22
每 100 万字符
语音转文字
  • Batch transcription：$0.18 · 批量
  • Fast transcription：$0.36 · 批量
  • MAI-Transcribe-2：$0.54 · 批量
  • Real-time transcription：$1.00 · 实时
每小时音频
免费额度
Free (F0): 5 audio hours of real-time STT and 0.5M neural TTS characters a month. Batch not included.
备注
Prices are East US pay-as-you-go from the Azure retail prices API (the pricing page renders them client-side). A "Fast Transcription Promo" meter at $0.10/h is listed; the page says a promo runs until Dec 31, 2026.
  1. 访问 Azure AI Speech ↗
  2. Speech pricing ↗
  3. Retail prices API (East US) ↗
  4. 核对于

Rev.ai

语音转文字
  • Reverb：$0.20 · 批量
  • Reverb Foreign Language：$0.30 · 批量
  • Whisper Large：$0.30 · 批量
每小时音频
免费额度
Free credits worth 5 hours of Reverb ASR.
备注
Rounded up to the nearest second, 15 second minimum.
  1. 访问 Rev.ai ↗
  2. Pricing ↗
  3. 核对于

ElevenLabs

文字转语音
  • Eleven v4：$80
  • Eleven v4 Turbo：$40
  • Eleven v3：$80
  • Multilingual v2：$80
  • Flash / Turbo v2.5：$40
每 100 万字符
语音转文字
  • Scribe v2：$0.22 · 批量
  • Scribe v2 Realtime：$0.39 · 实时
每小时音频
免费额度
Free plan: 10,000 v4 characters (~10 min) or 20,000 Flash/Turbo characters a month, plus 4h30m Scribe or 2h30m Scribe Realtime.
备注
API usage is billed in USD, not plan credits. Same per-unit rate on every plan.
  1. 访问 ElevenLabs ↗
  2. API pricing ↗
  3. 核对于

Deepgram

文字转语音
  • Aura-1：$15
  • Aura-2：$30
  • Flux TTS：$45
每 100 万字符
语音转文字
  • Nova-3 Monolingual：$0.258 · 批量
  • Nova-3 Multilingual：$0.312 · 批量
  • Whisper Large：$0.288 · 批量
  • Nova-3 Monolingual (streaming)：$0.288 · 实时
  • Flux English：$0.39 · 实时
每小时音频
免费额度
$200 free credit, no card required.
备注
Streaming rates are limited-time promo prices; regular prices shown in notes. Growth plan ($4K+/yr prepaid) is up to 20% less.
  1. 访问 Deepgram ↗
  2. Pricing ↗
  3. 核对于

OpenAI

文字转语音
  • tts-1：$15 2027-01-06 停用
  • tts-1-hd：$30 2027-01-06 停用
  • gpt-4o-mini-tts：$0.60 / 1M text in · $12 / 1M audio out (tokens) 2027-01-06 停用
每 100 万字符
语音转文字
  • gpt-transcribe：$0.27 · 批量
  • gpt-4o-mini-transcribe：$0.18 · 批量 2027-02-26 停用
  • gpt-4o-transcribe：$0.36 · 批量 2027-02-26 停用
  • whisper-1：$0.36 · 批量 2027-02-26 停用
  • gpt-live-transcribe：$1.02 · 实时
每小时音频
免费额度
未知 no free audio quota on the pricing page
备注
tts-1, tts-1-hd and gpt-4o-mini-tts shut down Jan 6, 2027; whisper-1 and the gpt-4o transcribe models shut down Feb 26, 2027 (deprecations page).
  1. 访问 OpenAI ↗
  2. API pricing ↗
  3. Deprecations ↗
  4. 核对于

Amazon (AWS)

文字转语音
  • Polly Standard：$4
  • Polly Neural：$16
  • Polly Generative：$30
  • Polly Long-Form：$100
每 100 万字符
语音转文字
  • Transcribe batch：$0.36 · 批量
  • Transcribe streaming：$0.60 · 实时
每小时音频
免费额度
Polly: 5M Standard characters a month; Neural 1M, Generative 100K, Long-Form 500K a month for 12 months. Transcribe: 60 min a month for 12 months.
备注
Transcribe prices are US East (N. Virginia), from the AWS price list API.
  1. 访问 Amazon (AWS) ↗
  2. Polly pricing ↗
  3. Transcribe pricing ↗
  4. Transcribe price list (us-east-1) ↗
  5. 核对于

Cartesia

文字转语音
  • Sonic-3.6：$65
每 100 万字符
语音转文字
  • Ink-2：$0.70 · 实时
每小时音频
免费额度
Free plan: 20K credits a month (~27 min of TTS or ~1h51m of STT).
备注
Credit-based plans: Pro $5 (100K), Startup $49 (1.25M), Scale $299 (8M). 1 credit = 1 TTS character; STT uses 3 credits per second. Overage: $65 (Pro), $45 (Startup), $38 (Scale) per 1M credits.
  1. 访问 Cartesia ↗
  2. Pricing ↗
  3. 核对于

未收录

  • PlayHT：Shut down: API offline July 2025, service closed Dec 31, 2025 after Meta acquired PlayAI. play.ht did not resolve on 2026-10-23.
  • LMNT：lmnt.com/pricing says "LMNT has shut down" (2026-10-23).
  • Hume：Pricing page renders client-side; no per-unit rates readable yet.
  • Mistral Voxtral：Speech models are priced per minute, but the rates are not on the public pricing page text.

邮件通知

额度一变，就通知你

套餐有变动时发一封简短邮件。不发垃圾邮件，我们也被限流。