Online help

How to use Speech Mic for live speech translation — from creating an account and adding credit to running a session with voice input and sharing a public listen link with your audience.

Overview

Speech Mic converts live speech into translated sentences in real time. A host runs a translation session (in your chosen source language), optionally using Voice Input for hands-free capture. Listeners open a public listen link in a browser — no app install — to read translations and optionally hear them spoken.

Create an account and (if enabled) add prepaid credit.
Create a translation — pick a source language from 74 options and one or more target languages (up to 10).
Speak or type source text; each completed sentence is translated automatically.
Publish to a listen collection and share the collection URL or QR code.
Listeners follow along on the public page with text, optional live audio, optional projection mode for a second screen, and feedback tools.

Video demo

Watch short walkthroughs of Speech Mic — creating a session, using voice input, and sharing a public listen link.

Accounts & sign-in

Create an account

Go to Create Account (/signup ) and enter your name and email.
You may check Sign up without password (passwordless login via email) to skip setting a password. You will receive a login link by email instead.
After you verify your email, new accounts may receive a one-time signup bonus of $20.00 in translation credit.

Sign in

Login (/login ) with email and password, or use Login by Email Link for passwordless access.
Use Forgot your password? if you need to reset a password.

Profile & access

Account (/account ) — update profile, change password, verify email, or delete your account.
The Translation menu requires a user role with translation access. If you do not see it after login, contact your site administrator.
Session timeout: you are logged out after a period of inactivity; stay signed in by using the app regularly during long events.

Credits & billing

Prepaid credits pay for translation voice input while the microphone is active in the editor. Charges accrue per minute of voice capture.

How charges are calculated

Base fee × number of target languages (each target language is billed per minute of voice input).

Public listeners are not billed. View your balance, active rates, and transaction history under Account in the top navigation (Account balance ).

Add credit (Stripe)

Open Account → Add credit (Stripe) .
Choose a pack amount and complete checkout securely.
Your balance updates immediately after payment.

While Voice Input is recording, the editor header shows an estimate of how many minutes your current balance can cover.

Create a translation session

From Translation → Create translation (/translations/new ):

Basic settings

Title — e.g. “Sunday service — English” (shown to listeners).
Source language — the language you speak or type. Use popular shortcuts or search on the New translation form (74 languages available; defaults to Chinese).
Target languages — pick one or more (up to 10) from 74 supported output languages. The first selected language is the Primary target (used for default text-to-speech and display). Search or use popular-language shortcuts. For Traditional Chinese output, add Chinese (Traditional) .
Domain — e.g. church/events vs general. Affects the default AI translation style (platform standard for that domain) and which voice hot words apply when voice input starts.

Voice input defaults (optional)

You can set defaults when creating the session; they can be changed later in the editor:

Voice input — microphone distance — Far-field — room / pulpit mic for a lectern or room mic; Near-field — lapel / headset for a clip-on or headset mic close to your mouth.
Voice input — Chinese recognition mode (Chinese source only) — Auto Chinese / English mix , Chinese-first, some English , or Pure Chinese .

Publish for listening

Check Publish for listening when creating, then choose a listen collection (see below). You can also publish later from the editor.
Submit Create and open to start the session.

Supported languages

Speech Mic supports 74 source languages and 74 target languages — including Chinese, English, Spanish, Japanese, French, Korean, and many more.

Source — what the host speaks or types in the editor. When voice input is enabled on your site, the microphone button works for any supported source language.
Target — what listeners read and hear on the public page. Choose up to 10 targets per session; the first is the primary line in the editor.
The source language cannot also be selected as a target in the same session (for example, you cannot use Spanish as both source and target).
On New translation , use Popular languages shortcuts or Search all languages for both source and target pickers.

Voice input (speech capture)

When voice input is enabled on your site, the editor shows a Voice Input (microphone) button. Speak naturally — recognized text appears in the source area, and each completed sentence is translated automatically.

For some source languages, text may appear a few seconds after you speak. Keep the editor tab visible for best results.

Starting voice input

Open your translation in the editor.
Allow microphone access when the browser prompts you.
Click Voice Input . Status shows Listening… when active.
Speak naturally. Text appears in the source area; each completed sentence triggers translation.
Click the stop button (same control) to end voice input.

Microphone distance (Voice Mic)

Far-field — room / pulpit mic — microphone farther away (pulpit, podium, room PA feed). Use when the mic is not close to your mouth.
Near-field — lapel / headset — mic on your body (lapel clip, headset). Use for close, consistent capture.
Set on the New translation form or in the editor under the source card ( Voice Mic dropdown). Changes apply the next time you start voice input.

Chinese recognition mode (CN Mode)

Available when source language is Chinese:

Auto Chinese / English mix — recognizes mixed Chinese and English words.
Chinese-first, some English — optimized for mostly Chinese with occasional English terms.
Pure Chinese — minimizes English recognition.

For non-Chinese source sessions, CN Mode is not applicable (shown as N/A).

Tips

The source textarea is read-only while voice input is active — use Voice Input to add text, or stop voice to edit manually.
Keep the editor tab visible when possible; locking the screen or backgrounding the browser can slow capture on some devices.
Credits are consumed only while voice input is actively recording (if billing is enabled).

Translation editor

Open a session from Translation → click a title. The editor has two main areas:

Source — text in your source language (typed or from voice input). Autosaves to the server.
Translations — one card per translated sentence, newest at the top.

During a live session

Complete a sentence with punctuation appropriate to your source language (e.g. 。, ？, ！ for Chinese; ., ?, ! for English and many other languages) so the system knows when to translate.
Character count shows how much source text has been translated vs total length.
Add language (editor header) — add another target language during a live session. New sentences translate into the added language from that point on; earlier sentences are not re-translated.
Reset (source card) clears all source text and segments — use with care.

Per-sentence actions (host)

Edit (pencil on a segment) — change the source sentence and Save & retranslate . This updates the segment list; it does not change live voice capture text.
Delete sentence — remove one segment.
Reported Issues — review flags submitted by listeners ( Dismiss when resolved).

Host playback (optional)

Live — automatically speak each new translation in time order (oldest → newest).
Play From Start — play all segments from the beginning.
Choose TTS voice and playback speed in the translations toolbar.

Your translation settings

On the Translation list page, toolbar buttons save settings on your account . They apply to translations you create (and use your personal defaults when you run a session). Other users on the same site keep their own settings.

Translation AI prompt

Opens the Translation AI prompt modal. Adjust how the AI translates (tone, terminology, formatting).
Choose a domain , then optionally set your personal override for that domain. If you leave the override empty, the system standard prompt for that domain applies (configured platform-wide by administrators).
Overrides are per domain , not one prompt for every domain. The same prompt body is used for every source-to-target pair; the app adds a source/target-language prefix when translating.
Use when you want a consistent personal style across many events (e.g. formal church language vs casual meetings).

Domain hot words

Your phrases that help voice recognition for a chosen domain (names, places, specialized terms). One phrase per line.
Merged with system hot words configured by administrators (your phrases are added on top).
Best for proper nouns and vocabulary that speech recognition often mishears.

Replacement vocabulary

Find/replace rules applied after speech recognition and before translation — fixes recurring mis-hearings (e.g. a name always wrong).
Longer “find” strings take priority over shorter ones.
Saved on your account; use for systematic corrections you want on every session you run, not one-off typos.

Listen collections

A listen collection gives listeners one stable URL for many translation sessions over time (e.g. “Sunday services” always at the same link).

Manage collections

Translation → Manage collections (/translations/collections ).
New collection — enter a title (e.g. “Main auditorium — live translation”).
Each collection gets a permanent public path:/public/translations/c/<collection-id>
Rename or delete collections from the same page. Deleting a collection does not delete your translation sessions.

Publish a session to a collection

When creating a translation, check Publish for listening and pick a collection.
In the editor header, toggle Published for listening and select or change the collection.
Published sessions appear on the collection list page for listeners.

Public listening page (for audiences)

Listeners open your shared link in Chrome, Safari, or Edge on phone, tablet, or computer. No account required.

Finding a session

Collection page — lists published sessions; a Live badge indicates recent activity.
Tap or click a session title to open the listen view.

Controls

Reading language — when multiple target languages are available, choose which translation to read and hear. Preference is remembered in your browser.
Live — queue new translations and speak them automatically in order (oldest → newest). Skips lines already present when you turn it on.
Text-to-speech voice — Microsoft Edge neural voice for the selected reading language.
Playback speed — 0.8× to 1.5×; remembered in your browser.
Source Text — show the original source sentence above each translation.
Text size (A − / A + ) — adjust reading size; remembered in your browser.
Replay All — play all segments from the oldest to the newest.
Sentence navigator (floating buttons) — jump to first, last, or current sentence; pause play-all.
Per sentence: Play (speaker icon) — play from that line forward in time order.
Flag (flag icon) — mark a possible translation issue for the host ( Mark possible translation issue / Remove translation issue flag ).
Corrected — if the host retranslated a sentence, view before/after corrections.

Projection mode (second screen / venue display)

Use Projection on the public listen page when you want a large, clean subtitle view on a projector, TV, or second monitor — for example at the front of a room while you operate audio on a laptop.

Projection (TV icon in the toolbar) — toggles projection on thesame page . The sentence list and normal controls hide; a full-screen view shows the line currently playing (dark background by default; in sync with Live , Replay All , or per-sentence Play ).
Projection toolbar — move the mouse to the top of the screen to show controls:
- Prev/Next (switch, off by default) — when on, shows the preceding and following sentences in faded text above and below the current line for a rolling subtitle effect. When off, only the current sentence is shown.
- A − / A + — text size for the projection view (remembered in your browser).
- Moon (moon icon) — toggle background and text between dark (black background, light text) and light (white background, dark text). Your choice is remembered in your browser.
- Fullscreen — expand to the whole display.
- Exit — leave projection mode (or press Esc ).
Source Text — if you turned on Source Text on the listen page, projection shows the original source line above the translation in each slot (current and, when Prev/Next is on, the faded adjacent lines too).
Two-tab setup — open the same listen URL in a second browser tab or window on the projector computer. Turn on Projection there and keep playback ( Live or Replay All ) on the operator tab. The projector tab follows the current line automatically (same browser, same site).
Left-aligned text is used in projection so long lines read naturally on a wide screen.

Realtime updates

New translations appear automatically while the host is speaking. A brief Realtime: connecting badge may show if the connection is re-establishing; the page catches up when back online.

Tips for listeners

Use headphones in quiet venues; tap the page once if the browser blocks automatic audio ( Live mode).
Collapse the header (chevron) for more reading space on small screens.
For a venue screen, use Projection mode (see above) — operator tab for audio, Projection on the display tab or monitor.
For help in another language for menus only, use UI Language / 界面语言 in the top bar (does not change the live translation text).

UI language (menus only)

The UI Language / 界面语言 control translates buttons and labels (e.g. Help, Login) using Google Translate in your browser. It does not change the live translation content — that is controlled by the host’s target languages and the listener’s Reading language picker.

Support, privacy & legal

Support — signed-in users can openSupport tickets for account or technical help.
Contact —support@speechmic.com
Privacy Policy andTerms of Use .
Privacy request — ask about personal data related to your account or public listen activity.

Troubleshooting

Voice Input button missing or disabled: Voice input must be enabled for your site. Contact your administrator if the microphone button never appears.
Microphone permission denied: Allow microphone access in browser settings for this site. On HTTPS only (or localhost for development).
Voice input active but no text appears: Keep the editor tab visible and speak in complete sentences. Stop and restart voice input if capture stalls. For some source languages, text may appear a few seconds after you speak.
Insufficient credit / voice stops: Add credit under Account . Voice input billing runs while recording; ensure balance covers your event length and target languages.
Speech playback disabled on listen page: The server may have TTS disabled. Listeners can still read text; contact the host or administrator.
Listeners do not see new translations: Confirm the session is Published for listening . Listeners should refresh or wait for realtime reconnect. Host should keep the editor tab open during the event.
Wrong language heard on listen page: Listeners should set Reading language to the desired target. Host should verify target languages on the session.
Projection screen not updating: For a two-tab setup, playback must run on the operator tab ( Live or Replay All ) in thesame browser as the projection tab. Refresh the projection tab and turn Projection on again if needed.
Translation menu not visible after login: Your account needs translation access. Contact your site administrator to enable the translation role.
Locked or background browser tab: Host: avoid locking the PC during voice capture when possible. Listener: return to the tab to refresh; live updates resume when visible.

Still need help? Sign in and use Support , or email support@speechmic.com .