Skip to content

Catalogaudio

CosyVoice

situationalBlueshift — gaining momentum

Text-to-speech and voice-cloning model family from Alibaba Cloud.

Visit github.com →Add to kitOpen in Noemium

ledger entry · observed by @whysanesanders

Best-in-class Chinese TTS naturalness, plus open-source weights you can self-host. Pricing is token-based and can be confusing compared with per-minute audio APIs; English quality is good but not market-leading.

Strengths

  • Top-tier naturalness for Chinese text-to-speech.
  • Open-source weights that can be self-hosted.
  • API and voice-creation options on Alibaba Cloud.

Known limitations

  • Token/character pricing can be confusing vs per-minute audio APIs.
  • Commercial rights and voice-cloning ethics need review.
  • English and non-Chinese quality are good but not market-leading.
  • Same Alibaba Cloud region/compliance friction as other Qwen services.

Use it when

  • Chinese-language applications that need natural-sounding TTS.
  • Teams that can self-host to keep voice data off shared cloud endpoints.

Skip it when

  • You need simple per-minute pricing and predictable audio costs.
  • English or non-Chinese output must be market-leading.
  • Alibaba Cloud region and compliance requirements are a blocker.

Buy-check

  • ·Price — paid, CosyVoice 2 from ~$0.287/10,000 characters; CosyVoice 3.5 Plus from ~$0.22/10,000 characters; CosyVoice voice creation free; Qwen-TTS voice creation $0.20/voice; free quota: 10 voice creations in first 90 days (Singapore region) as of 2026-08-24. Open the pricing page, not a screenshot.
  • ·Free tier — Free tier exists — burn it first.
  • ·Cancel fallout — What happens to your data/projects on cancel — check the ToS, not the FAQ.
  • ·Seat math — Per-seat pricing multiplies quietly. Count seats before the annual plan.
  • ·Who skips it — Also skipped when: You need simple per-minute pricing and predictable audio costs.; English or non-Chinese output must be market-leading..

Price trail — 2 revisionschangelog →

  • 2026-W35CosyVoice 2 from ~$0.287/M input tokens; CosyVoice 3.…CosyVoice 2 from ~$0.287/10,000 characters; CosyVoice…
  • 2026-W34CosyVoice 2 from ~$0.287/M input tokens; voice creati…CosyVoice 2 from ~$0.287/M input tokens; CosyVoice 3.…

Facts

pricing
paid1
price note
CosyVoice 2 from ~$0.287/10,000 characters; CosyVoice 3.5 Plus from ~$0.22/10,000 characters; CosyVoice voice creation free; Qwen-TTS voice creation $0.20/voice; free quota: 10 voice creations in first 90 days (Singapore region)
free tier
yes — paid to keep using
open source
yes
api
yes
self-host
yes
category
audio
evidence
source-verified
Models
locked

Data: trains on inputs unknown · local processing yes

Sources

  1. 1github.com/QwenAudio/CosyVoice[snapshot]
  2. 2alibabacloud.com/help/en/model-studio/voice-design-user-guide[snapshot]
  3. 3cloudprice.net/models/alibaba-cosyvoice-2[snapshot]
  4. 4cloudprice.net/models/alibaba-cosyvoice-3-5-plus[snapshot]
  5. 5guijiagi.com/posts/ai-tts-2026-comparison/[snapshot]

No source, no number.

Edit this page on GitHub

Similar joball alternatives →

toolverdictpriceossself-host
CosyVoicesituationalpaidyesyes
Deepgramshipfreemiumnono
ElevenLabsshipfreemiumnono
Sunoshipfreemiumnono

Facts from the catalog files — not scores you can buy. Quality and speed stay in the briefing, not in a fake 1–5 grid.

Deepgram
audioship

Deepgram

Speech-to-text API built for developers — fast, cheap, real-time.

APIFREE TIER

freemium$200 free credits; Nova-3 STT from $0.0048/min (promo)

Steady momentum
ElevenLabs
audioship

ElevenLabsanchor

The reference standard for AI voice — TTS, cloning, dubbing and voice agents.

APIFREE TIER

freemiumfree tier (10k credits/mo, non-commercial); Starter $6/mo, Creator $22/mo

Blueshift — gaining momentum
Suno
audioship

Suno

Full-song generation — vocals, lyrics, arrangement — from a text prompt.

FREE TIER

freemiumfree tier (10 songs/day, non-commercial); Pro $10/mo, Premier $30/mo

Blueshift — gaining momentum