Voice input lets you dictate notes, emails, and AI commands using your voice instead of typing. Powered by OpenAI gpt-4o-mini-transcribe (Whisper), it handles accents, background noise, and adds punctuation automatically. Available in the AI chat, note composer, and email composer.
Whether you are capturing meeting notes immediately after a client call, composing a lengthy follow-up email, or issuing a quick command to the AI assistant, voice input eliminates the friction of typing on both desktop and mobile devices. This guide covers every aspect of the feature, from activating the microphone to editing transcriptions and getting the best possible accuracy.
Where Voice Input Is Available
The microphone button appears in three locations throughout SalesSheet, each tailored to a different workflow:
- AI Chat Input: Dictate commands to the AI assistant such as creating contacts, drafting emails, or searching your CRM. The transcribed text is sent to the AI exactly as if you had typed it.
- Note Composer: When adding a note to a contact, deal, or activity record, the microphone icon appears on the right side of the note text area. This is ideal for dictating meeting summaries and call debriefs.
- Email Body Editor: While composing an email within SalesSheet, you can dictate the entire body of the message. The transcription flows into the rich text editor with proper punctuation and paragraph breaks.
In all three locations, the microphone icon sits on the right side of the input field. It is a small circular button with a microphone glyph. If you do not see the button, check that your browser has granted microphone permissions to the SalesSheet domain (see the Troubleshooting section below).
How to Activate Voice Input
Find the Microphone Button
Look for the microphone icon on the right side of the input field. It appears in three places: the AI chat input, the note composer, and the email body editor. The button is a small circle with a microphone icon. On mobile devices, the microphone button is slightly larger for easier tapping and sits in the same position relative to the input field.
Grant Browser Permissions
The first time you click the microphone button, your browser will display a permission prompt asking you to allow microphone access for the SalesSheet domain. Click "Allow" to proceed. This permission only needs to be granted once per browser. If you accidentally deny permission, you can change this in your browser settings by clicking the lock or site settings icon in the address bar.
Start Recording
Click the microphone button to start recording. The button pulses red to indicate it is actively listening. A waveform visualization appears above the input to provide visual feedback that audio is being captured. Speak naturally at a conversational pace. Whisper handles pauses, filler words, diverse accents, and background noise with high accuracy. There is no strict time limit on a single recording, but for best results keep individual recordings under two minutes and use multiple recordings for longer content.
Review the Transcription
Click the microphone button again to stop recording. Your spoken words appear as text in the input field within one to three seconds, depending on the length of the recording. The transcription includes proper punctuation, capitalization, and paragraph breaks. Review the text carefully and make any edits before sending or saving. You can click the microphone again to append additional dictation to the existing text without overwriting what is already in the field.
Send or Save
Once you are satisfied with the transcription, hit Enter to send (in AI chat) or click Save/Send as appropriate. Voice input works identically to typed input. The AI assistant processes voice-transcribed commands the same way it handles typed ones, and notes or emails created via voice are stored and displayed without any distinction from manually typed content.
Supported Fields and Use Cases
Voice input supports the following field types and scenarios within SalesSheet:
- Plain text notes: Contact notes, deal notes, and activity notes all accept voice input. The transcription preserves natural paragraph structure.
- Rich text email bodies: When dictating into the email composer, the transcription respects paragraph breaks. You can further format the text using the rich text toolbar after dictation.
- AI chat commands: Any command you can type into the AI chat can be dictated instead. This includes contact creation, email drafting, deal updates, search queries, and scheduling tasks.
- Subject lines: You can click into the email subject field and use voice input there as well, though shorter phrases work best for subject lines.
Nota
Voice input is not currently available in custom form fields or bulk edit screens. It is supported only in the AI chat, note composer, email body editor, and email subject line fields.
Tips for Better Accuracy
The Whisper transcription engine delivers strong accuracy out of the box, but the following practices will help you get the best results:
- Use an external microphone or headset. Built-in laptop microphones pick up fan noise, keyboard sounds, and room echo. A dedicated microphone or headset significantly improves transcription quality.
- Speak at a natural pace. You do not need to speak slowly or over-enunciate. Whisper is trained on conversational speech and handles normal speaking cadence well.
- Minimize background noise. While Whisper can filter out moderate background noise, loud environments with overlapping conversations will reduce accuracy. Close the door or move to a quieter space when possible.
- Dictate punctuation for precision. Whisper adds punctuation automatically, but if you need a specific structure you can say "period," "comma," or "new paragraph" and the engine will interpret these correctly in most cases.
- Keep recordings under two minutes. For longer content, pause and click the microphone again to start a new segment. This reduces the chance of the engine losing context over very long recordings.
- Spell out uncommon proper nouns. Industry-specific jargon and uncommon company names may occasionally be transcribed incorrectly. If you notice a repeated error, spell the word out or manually edit the transcription after dictation.
Editing Transcriptions
After the transcription appears in the input field, you have full editing control before sending or saving. You can click anywhere in the transcribed text to place your cursor, select words to replace, or use standard keyboard shortcuts to cut, copy, and paste. If you need to add more content via voice, simply click the microphone button again and speak. The new transcription will be appended at the current cursor position rather than replacing the existing text.
For email bodies, the rich text toolbar remains fully functional after voice input. You can bold key phrases, add bullet points, insert links, or restructure paragraphs using the formatting controls. Voice input provides the raw text; formatting is entirely up to you afterward.
Language Support
SalesSheet's voice input leverages OpenAI Whisper, which supports transcription in over 50 languages. The engine automatically detects the language you are speaking, so there is no need to manually select a language setting. If you frequently dictate in a language other than English, Whisper will adapt its punctuation and formatting rules accordingly. Mixing languages within a single recording (for example, using a Spanish company name in an otherwise English sentence) is handled well in most cases, though very rapid switching between languages may reduce accuracy slightly.
Consejo Pro
Voice input is perfect after client calls. While the conversation is fresh, click the mic and dictate your meeting notes. You will capture details you might forget if you wait to type them out later. The AI adds punctuation automatically, so just speak naturally without worrying about formatting. Many users find that a 30-second voice note right after a meeting captures more useful detail than five minutes of typing an hour later.
Solución de Problemas
The microphone button does not appear.
Voice input requires browser microphone permissions. Check that you have allowed microphone access for the SalesSheet domain. In Chrome, click the lock icon in the address bar and verify microphone is set to "Allow." In Firefox, go to Settings > Privacy and Security > Permissions > Microphone and ensure SalesSheet is listed. If you previously blocked microphone access, you will need to manually change it to "Allow" and then refresh the page.
Transcription quality is poor or words are missing.
Speak clearly and at a normal pace. Avoid very noisy environments. If quality is consistently poor, check your microphone hardware. Built-in laptop microphones can pick up fan noise. An external microphone or headset typically produces much better results. Also verify that the correct input device is selected in your operating system's audio settings, especially if you have multiple microphones connected.
There is a long delay before the transcription appears.
Transcription typically completes within one to three seconds. If you experience longer delays, check your internet connection. The audio is sent to the transcription service in the cloud, so a slow or unstable connection can increase processing time. Very long recordings (over two minutes) will also take longer to process. Try shorter recording segments for faster results.
The transcription replaced text I already had in the field.
Voice input appends text at the current cursor position. If you had text selected when you started recording, the transcription will replace the selection. To avoid this, click to place your cursor at the end of the existing text (without selecting anything) before starting a new voice recording.