免费在线 AI 语音：通话 · 转写 · 朗读，同一页完成
NorSpeak is a free aivoice online workbench. Start free aivoice in the browser: talk in realtime, get speech-to-text, and run free AI voice TTS—no install. This free AI voice page puts free aivoice calls, transcription, and readout in one place so you can try free aivoice online in under a minute.
Choose TTS, STT, or Live. Your current tool stays on this page.
NorSpeak
Free AI voice workbench
Turn a script into natural audio in one request.
为什么在 NorSpeak 体验免费 AI 语音
Voice tools are easier to judge when listening, speaking, and editing happen together. NorSpeak gives each task a clear place while keeping the session on one page.
A voice demo should answer practical questions quickly. Does the reader pronounce a product name correctly? Can a live assistant respond before the pause feels awkward? Does a recording become text you can actually edit? NorSpeak lets you test those questions without preparing a local environment or moving files across separate utilities. Paste a script, choose a real supported language, and play the result. Open Live when you want a conversation. Use Dictation when the words matter more than the performance.
The shared workspace also makes comparison honest. Text-to-speech playback returns to the same player where the audio can be downloaded. Speech-to-text results stay available for correction and export. A realtime session keeps user and assistant captions in a readable timeline, alongside latency measurements when the provider exposes them. You can hear the experience, inspect the transcript, and decide what belongs in your product before writing integration code.
Move from a written line to audio, a recording to text, or a live exchange without learning another interface.
Download synthesized audio, export the transcript, and preserve a completed session instead of replaying a request.
Use the browser at no charge, then add paid API credit only when an external product needs it.
朗读、转写、实时对话，各自清楚又彼此相连
Start with the job in front of you. Every mode is designed to finish a useful piece of work, not just show a model response.
Paste a script, select a supported language and voice, then listen in the page. The completed audio remains available to download in the format returned by the service.
Use it for product narration, lesson drafts, accessibility checks, pronunciation review, or a fast voiceover reference. Because download reuses the synthesized result, exporting does not spend time or credit on a duplicate request.
Connect, speak naturally, and follow both sides in a caption timeline. The center orb communicates listening, thinking, speaking, and interruption states without covering the conversation.
Full duplex supports a fluid exchange. Push-to-talk remains useful in noisy rooms. When a user interrupts playback, the visual response is connected to the real cancel event so the interface reflects what the session actually did.
Record, dictate, or upload audio and receive text in the workspace. Review the words while the source is still close at hand, then export a clean transcript for the next step.
This is useful for interviews, meeting notes, rough ideas, support calls, and subtitle preparation. The goal is a text result you can use, not a transcript locked inside a temporary demo.
These tools fit a natural loop. A creator can dictate an outline, edit the transcript, and send the final paragraph through the reader. A support lead can test a spoken response, run a practice call, and keep the captions for review. A developer can compare browser behavior with an API response using the same input. Each step is independent, but the shared page removes the friction of signing into unrelated services or rebuilding context.
工作室处理声音，听写模式专注文字
The interface separates expressive voice work from focused transcription while keeping both close enough to move between them.
Studio brings text-to-speech and Live together. It is the place to shape how a line sounds, test timing, or have a responsive voice conversation with captions.
Dictation reduces the interface to capture, transcription, correction, and export. Use it when clean text is the deliverable and the voice is the source.
Studio is intentionally product-like rather than a wall of model settings. The main view centers the action: start playback or connect a call, then watch and listen. Model, voice, transcription bias, and inference controls live in Advanced for people who need them. This keeps a first session understandable while preserving the controls required for careful evaluation.
Dictation follows a quieter rhythm. Record or choose a file, wait for the transcript, review uncertain phrases, and export. It avoids ornamental controls and keeps the source associated with the text. When the transcript becomes a script, move it back to Studio and hear the revised version. The two modes form a practical editing cycle without pretending that every voice task needs the same interface.
从允许麦克风到导出结果，只需一个清楚的流程
A session is short enough to try immediately and structured enough to produce something worth keeping.
Open TTS for a written script, STT for recorded speech, or Live for a two-way conversation. The active tab preserves the context needed for that job.
Microphone permission is requested for recording or Live. Text playback does not need it. A file stays part of the chosen transcription request.
Play the generated audio, read streaming captions, or correct a transcript. Live shows response state and measured timing without crowding the primary controls.
Save the returned audio without synthesizing again. After Live, export the transcript and available recording with accurate file extensions and MIME types.
For text-to-speech, the shortest path is paste, choose, play, download. The audio response becomes the current result and remains the source for both playback and export. If the service returns MP3, the file uses MP3. If it returns WAV, OGG, or WebM, NorSpeak keeps the matching type and extension. The download action does not make a hidden second synthesis request.
For Live, connecting moves attention to the voice state rather than a settings form. The orb begins idle, passes through connecting, and then responds to microphone and playback levels. Listening, thinking, and speaking remain visually distinct. Captions build in a left and right timeline so the user and assistant can be scanned separately. When timing data exists, a compact HUD reports time to first response and first audio.
Barge-in is part of the real interaction. If the user speaks while assistant audio is playing and the session cancels that audio, the orb flashes an interrupt ripple. The effect does not run for an ordinary microphone change. At the end, a single export action collects the transcript and the recording that actually exists. That makes the session useful for review, support notes, interviews, and product testing.
页面内体验免费，外部调用使用付费额度
NorSpeak separates trying the product from operating an integration. The page is open for hands-on use. API keys and credit belong to software calling the service outside the page.
Open the browser tools for TTS, STT, and realtime conversation. Test a script, microphone, file, or live flow before deciding whether an integration is necessary.
Open the workbenchChoose a subscription or one-time credit pack when a server, worker, automation, or customer-facing app needs to make external requests with a NorSpeak API key.
View API pricingThis boundary keeps the first experience simple. Someone evaluating a voice can work on the page without choosing a billing plan first. Someone shipping an application gets a key, usage accounting, and a clear paid relationship for calls made by that application. The distinction also prevents marketing language from implying that unlimited third-party API traffic is included in a browser demo.
按使用节奏选择订阅，或一次性补充额度
The online workbench remains available for direct use. These plans fund API calls made by your own applications, automations, and services.
$9.90
per month
A flexible plan for active personal projects and prototypes.
$24.90
every 3 months
Lower monthly cost for products moving through a build cycle.
$79.00
per year
The best rate for ongoing voice workflows and integrations.
one time
$4.99
Top up a small test, demo, or one-off production run.
one time
$12.99
Extra credit for a launch sprint or a larger batch of audio.
one time
$29.99
A larger reserve without starting another subscription.
Checkout, credits, renewals, and subscription status are handled through the NorSpeak account. Taxes may apply where required.
关于免费使用、开始方式与 API 计费
The short answers below are also used for the page FAQ structured data, so search metadata and visible guidance stay aligned.
Free aivoice means browser-based AI voice: talk, transcribe, and speak text aloud without installing an app. NorSpeak is a free aivoice online workbench.
Yes. On-page free aivoice (realtime, STT, TTS) is free; API keys for external calls are paid.
Open norspeak.com (or /free-aivoice), allow the mic, and start a free aivoice session in one click.
现在就打开浏览器里的 NorSpeak 工作台
Paste a line, bring a recording, or allow the microphone for Live. You can hear, read, and export a useful result before deciding whether your product needs the API.