Start with an article you meant to finish
Paste Chinese or English text, or import a public WeChat article. After email verification and password setup, pasting a whole article or importing a link can automatically prepare written key points. You can also submit typed text.
Review the key points and choose Listen to key points to generate speech. Choose the original when you need context. Pasting text does not automatically generate or play audio.
Listen at your own pace
Enjoy natural, fluent AI narration. Adjust speed, seek within a track and download completed MP3s. Replaying an existing track does not spend voice credits again. The site currently uses one configured voice; a voice selector is not available.
Switch the interface between Chinese and English. Summaries and speech follow the source language; changing the interface does not translate your article. AI key points can omit or misunderstand details. Check important information against the original.
The AI models behind Tingjian
Written key points use DeepSeek's DeepSeek-V4.1-Flash-1017. Speech synthesis uses Alibaba Cloud's Qwen Qwen-Audio-3.1-TTS-Flash. Key points and audio are AI-generated; verify important information.
The text request model is deepseek-flash, a rolling alias maintained by the provider. As of November 2, 2026, its official documentation identifies DeepSeek-V4.1-Flash-1017 as the serving model. The speech request model is qwen-audio-3.1-tts-flash. Providers may update alias targets; this disclosure will be updated when the site's configuration changes.
Tingjian is an independent service, not an official application of these providers or a claim of their endorsement. It currently uses fixed models and a configured voice, without a user model selector or voice-cloning feature.
Finish one article. Move on to the next.
Your account keeps articles, key points, audio and playback progress, so you can resume across devices. Use Next article to replace your input and Recent listening to find unexpired articles.
Content is kept for seven days from article creation. Cloud access stops at expiry and cleanup follows. Download audio you want to keep; downloaded files are under your control.
Know the current limits
Guests can try a fixed example. Personal generation requires an email account. Written key points have three free daily uses; free speech and generation also have account, shared-network and site-wide limits.
Full-article speech supports up to 5,000 characters. Written summaries support source text up to 20,000 characters. For longer articles, prepare written key points and listen to those. Link import supports publicly accessible WeChat articles; paste the text if the source blocks import.
Credit packs are one-time purchases without auto-renewal. Verified real payments grant the purchased pack, valid for 90 days. Orders explicitly marked as test mode do not charge real money and grant only 1 usable voice credit per successful test order. Speech retains a trial allowance safeguard and stops new generation when the trial is exhausted or expires. The current safeguard date is December 21, 2026.