Speak Input - Voice Dictation for ACF
| 开发者 | cmoraleda |
|---|---|
| 更新时间 | 2026年9月15日 00:48 |
| PHP版本: | 8.2 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
详情介绍:
acf_form().
Click the microphone, speak, click again to stop. The recording is sent to your chosen speech-to-text engine and the transcript is inserted into the field a moment later — nothing is transcribed live while you speak, and nothing leaves your site until you stop recording.
Free
- Voice dictation for
textandtextareafields. - Deepgram as the transcription engine, using your own free Deepgram API key.
- A dedicated "Speak Input" tab in every ACF field's settings, so you choose exactly which fields get a microphone button — off by default.
- Keyterm prompting: give Deepgram a list of names, jargon or product terms it should listen for.
- Dictation language follows the editor's browser language automatically.
- Works in the admin and in front-end forms built with
acf_form(), including for anonymous visitors, with a configurable request limit to protect your API usage.
- Dictation for
wysiwyg(rich text) fields, in addition totextandtextarea. - A second transcription engine, OpenAI, selectable alongside Deepgram.
- Insertion modes per field: replace the whole field, append, insert at the cursor, or replace only the current selection.
- A configurable recording time limit per field.
- A site-wide and per-field context prompt (used by the OpenAI engine) to improve accuracy for names, jargon or the topic of your site.
- A fixed dictation language per field, with automatic WPML awareness.
- A diagnostics screen: a log of transcription attempts (engine, result, timing) to help you troubleshoot and, if needed, share with support — never your dictated text or audio, only metadata.
- Priority email support.
- Text
- Textarea
- WYSIWYG (rich text editor) — Pro
- Deepgram — Free and Pro. Requires your own Deepgram API key (a free tier is available).
- OpenAI (
gpt-transcribe) — Pro only. Requires your own OpenAI API key.
安装:
- Make sure Advanced Custom Fields (free or PRO) is installed and active.
- Upload the
speak-input-acffolder to/wp-content/plugins/, or install the plugin through the Plugins screen in WordPress. - Activate the plugin through the "Plugins" screen.
- Go to Settings > Speak Input ACF and add your Deepgram API key (a free account and starter credit are available at deepgram.com).
- Edit any ACF field group. Open a
textortextareafield's settings and switch to the new "Speak Input" tab. Turn on "Enable voice dictation" for that field. - Save the field group. A microphone button now appears on that field wherever it's used — in the admin, and in any front-end form built with
acf_form(). - (Optional) Upgrade to Pro at speakinput.com/advanced-custom-fields for
wysiwygfield support, the OpenAI engine, insertion modes, and the rest of the Pro features listed above.
屏幕截图:
常见问题:
Do I need my own API key?
Yes. Speak Input doesn't include or resell transcription — you sign up for your own Deepgram account (a free tier is available) and, on Pro, optionally an OpenAI account, then enter your API key in Settings > Speak Input ACF. Your key is stored on your own site and never exposed to visitors.
Does this need ACF PRO, or does the free version of Advanced Custom Fields work?
The free version of Advanced Custom Fields is enough for dictation on text and textarea fields. Dictation on wysiwyg fields is a Speak Input Pro feature, independent of whether you're running free ACF or ACF PRO.
Is "Speak Input Pro" the same as "Advanced Custom Fields PRO"?
No — they're two separate products from two separate companies. "Speak Input Pro" is this plugin's own paid tier; "Advanced Custom Fields PRO" is a separate upgrade sold by the makers of ACF. You can combine either version of Speak Input with either version of ACF.
Is my dictated audio or text stored anywhere?
No. Audio is sent to your chosen transcription engine (Deepgram or, on Pro, OpenAI) only to produce a transcript, and is not stored by this plugin. The optional Pro diagnostics log records only metadata about each attempt (engine, result, timing) — never the dictated text or audio. See the "Privacy" section above and the "External Services" section below for details.
Does this work on the front-end, not just the WordPress admin?
Yes. Speak Input works with any acf_form() front-end form, including for visitors who aren't logged in. To protect your API usage from abuse, requests from visitors without edit access are limited (20 requests per 10 minutes by default); both the request count and the time window are configurable in Settings > Speak Input ACF.
Does "Fixed language" (or "Automatic") force dictation into that language?
No. The language setting — whether it's "Automatic" or a specific "Fixed language" you choose for a field — is a hint that helps the speech-to-text engine recognize your speech more accurately. It is not a translator and not a filter. If you dictate in a language different from the one configured, most of the time the engine will still transcribe what you actually said, in the language you actually said it, rather than translating it or refusing to transcribe it. Setting a field to English does not mean speech in another language gets translated into English, and it does not prevent you from dictating in that other language.
Which browsers are supported?
Speak Input relies on your browser's own microphone and recording APIs (getUserMedia and MediaRecorder), available in current versions of Chrome, Firefox and Edge. Safari's support for these APIs has historically been less consistent, so results there can vary by version.
Can I choose which fields show the microphone button?
Yes. Voice dictation is off by default on every field. Turn it on per field from that field's own "Speak Input" settings tab in ACF.
更新日志:
- Addressed WordPress.org Plugin Check warnings: removed the now-redundant manual translation loading (WordPress already auto-loads this plugin's translations on WordPress.org), documented the intentional direct database query used only during opt-in uninstall cleanup, and renamed internal activation/deactivation/bootstrap functions to consistently start with the plugin's prefix. No functional change.
- Fixed a fatal error in the free version: diagnostics logging (a Pro-only feature, physically excluded from the free package) was referenced without a safety check on plugin activation/deactivation, on every page load, and on every dictation request, crashing the free plugin. No change for Pro users.
- The diagnostics screen's JS asset is now also physically excluded from the free package, alongside the rest of the diagnostics feature.
- The plugin's text domain now matches its WordPress.org slug (
speak-input-voice-dictation-for-acf). - All Premium-only functionality — the OpenAI engine, wysiwyg support, insertion modes, recording time limit, prompts, and language controls — is now physically excluded from the free package (no change for Pro users).
- Renamed internal option, table and hook names to avoid conflicts with other plugins.
- The Pro-only Diagnostics feature is no longer bundled with the free version (no change for Pro users).
- Added
composer.jsonto the package and removed bundled translation files — translations are now managed through WordPress.org.
- Security hardening: added direct file access protection to two internal template files.
- Initial release.
- Voice dictation for
textandtextareafields, with Deepgram as the transcription engine. - Per-field "Speak Input" settings tab and global Deepgram keyterm prompting.
- Works in the admin and in front-end
acf_form()forms, including for anonymous visitors, with a configurable request limit. - Pro:
wysiwygfield support, OpenAI as a second engine, insertion modes, recording time limit, context prompts, fixed/WPML-aware language per field, and a diagnostics log.