# How to send AI voice notes with ElevenLabs Source: https://wassuply.com/docs/voice-notes/ Updated: 2026-10-30 · Wassuply 1.0.4 for Windows and macOS > Wassuply turns a written script into a separate AI voice note for every contact, using your own ElevenLabs API key, and sends it from your numbers as a normal WhatsApp voice message. The Voice Library manages your ElevenLabs voices: clone your own voice from audio samples, add voices from ElevenLabs' library and tune the speaking style. Wassuply has no microphone recorder of its own. In the app: Sidebar → Content → Voice Library ## Key facts - Wassuply generates one ElevenLabs voice note per contact after merge fields such as {name} and spintax are filled in, so each contact hears their own version of the script. - Wassuply sends AI voice notes as WhatsApp voice messages recorded through WhatsApp's own voice button, and never falls back to sending the audio as a file. - Wassuply records at most 180 seconds (3 minutes) of each voice note, so anything longer is cut off. - Wassuply's default ElevenLabs model is multilingual v2, with turbo v2.5 available as a faster option in the Voice Library. - Voice cloning in Wassuply accepts mp3, wav, m4a, ogg, flac and webm samples and suggests 1 to 3 minutes of clean, single-speaker audio. - Before a Wassuply voice campaign sends anything, it checks your ElevenLabs key, the chosen voice and, where it can read it, your remaining character credit. ## What are voice notes in Wassuply? A voice note campaign sends each contact a spoken message instead of text, generated by ElevenLabs from a script you write, in a voice you choose. - You need your own ElevenLabs API key, and ElevenLabs bills you for the characters you generate. - **Voice note** is a message type in the campaign wizards, so it works in [cold outreach](https://wassuply.com/docs/cold-outreach/) and [Broadcast Campaigns](https://wassuply.com/docs/broadcast-campaigns/). - The **Voice Library** (sidebar **Content → Voice Library**) manages the voices on your ElevenLabs account. - Wassuply cannot record from your microphone or send audio files you upload. To use your own voice, clone it. ## How do I add my ElevenLabs key? Paste your key in **Settings → Voice Notes** and save; the Voice Library then loads your voices. **How to connect ElevenLabs to Wassuply** 1. **Open the settings**: Go to **Settings → Voice Notes**. 2. **Paste the key**: Paste it into **ElevenLabs API key**. **Get an API key at elevenlabs.io** opens the ElevenLabs key page. 3. **Save**: Click **Save Changes**. 4. **Check it works**: Open the **Voice Library**. Your voices appear under **Your Voices**; a bad or missing key shows **Add your ElevenLabs API key** with the error. The key is stored on this computer and sent only to ElevenLabs, when Wassuply lists voices, clones a voice or generates audio. It is separate from your text AI key in [AI features](https://wassuply.com/docs/ai-features/). ## How do I manage voices in the Voice Library? The Voice Library has four tabs, and every change is made on your ElevenLabs account. - **Your Voices**: Every voice on your account, with a play button for its preview and **Refresh**. Cloned and added voices can be deleted, which removes them from ElevenLabs and cannot be undone. Premade voices cannot be deleted here. - **Style & Test**: How the voice is delivered, plus a two-voice comparison. - **Clone a Voice**: Creates a voice from your own audio samples. - **Voice Library**: Searches ElevenLabs' public voices by name, accent or style, filtered by **Any**, **Male** or **Female**, up to 100 per search. **Add to my voices** copies one to your account. There are no folders or tags. The list in **Your Voices** is your ElevenLabs account's voice list, and it is what the campaign wizard offers. ## How do I clone my own voice? Upload a few clean recordings of the voice on the **Clone a Voice** tab and ElevenLabs creates a new voice you can use in campaigns. **How to clone a voice in Wassuply** 1. **Record samples**: Use 1 to 3 minutes of clean audio with one speaker and no background noise or music. A few varied clips work better than one long clip, and a phone voice memo is fine. 2. **Name it**: Enter a **Voice name** and an optional **Description**. 3. **Add the files**: Click **Choose audio file(s)** and pick mp3, wav, m4a, ogg, flac or webm files. 4. **Confirm consent**: Tick the box confirming it is your own voice or you have the speaker's explicit consent. 5. **Clone**: Click **Clone Voice**. The new voice appears under **Your Voices**, ready for campaigns. Your samples are uploaded to ElevenLabs to build the voice, and you are responsible for how a cloned voice is used. ## How do I change how the voice sounds? Use the **Style & Test** tab. Changes save automatically and apply to every voice note, including running campaigns. | Setting | Default | What it changes | | --- | --- | --- | | Preset | Natural & friendly (recommended) | Sets emotion and expressiveness together. Also Expressive & energetic, Balanced, Calm & steady (professional) and Custom. | | Emotion & variety | 60% | Higher gives more intonation and less monotone delivery. | | Expressiveness | 35% | How animated the delivery is. | | Voice match | 80% | How closely it keeps the chosen voice's timbre. | | Boost clarity | On | Sharper audio. | | Model | Natural (multilingual v2) | The Fast option (turbo v2.5) is quicker and a little less rich. | Write scripts the way you speak, and add `...` or `` for pauses. The **A/B test** box speaks one line in two voices so you can compare them with **Generate & compare**; this uses ElevenLabs credits. ## How do I send voice notes in a campaign? Choose **Voice note** as the message type, pick a voice and write the script; Wassuply then generates and sends a separate clip for each contact as the campaign runs. - In the Script (or Message) step, choose **Write Message**, then **Voice note**. Voice notes cannot be combined with an attachment or with **Use Sequence**. - Pick a voice from the list, or click **More voices** to open the Voice Library. - Write the **Voice Script**. `{name}`, `{phone}`, `{custom1}`, `{custom2}` and spintax are filled in per contact before the audio is generated, so `Hi {name}` speaks each contact's name. - Click **Preview** to hear the first contact's version, or use **Send Test Message** to send a real voice note to a number you choose. - Launch as normal. The live log shows each voice note being generated and sent. If the key is invalid, the voice no longer exists or ElevenLabs reports no credit left on a plan that cannot go over its limit, the campaign is marked failed before anything is sent and the log says why. Follow-ups from a sequence are never voice notes. ## How are voice notes delivered? Each voice note arrives as a normal WhatsApp voice message with the waveform player, not as an audio file attachment. - ElevenLabs returns an MP3 for each contact. Wassuply keeps it in a temporary folder and deletes it after the send; leftovers older than 24 hours are cleared when the app starts. - Wassuply plays that audio into WhatsApp Web's own voice-message recorder in the chat, waits for the clip to finish, then sends it. - If WhatsApp's voice button is not found or recording does not start, that contact is marked failed with a voice note error. Wassuply never sends the audio as a document instead. - The message footer from **Settings → Messaging** is not added to voice notes. ## What are the limits and costs? Voice notes are capped at 3 minutes each, take longer to send than text, and cost ElevenLabs characters for every contact. | Limit | Value | | --- | --- | | Longest voice note recorded | 180 seconds; anything longer is cut off | | Time per contact | Generation (times out after 90 seconds) plus the length of the clip | | Audio generations | One per contact, plus one per Preview, test send or A/B test voice | | Clone sample formats | mp3, wav, m4a, ogg, flac, webm | | Voices per library search | Up to 100 | ElevenLabs bills the characters you generate on your plan; Wassuply does not charge for voice notes. Keep scripts short: the character counter turns amber at 300 and red at 500 characters. Your normal delays still apply between contacts, see [sending safety](https://wassuply.com/docs/sending-safety/). ## Frequently asked questions ### Can I upload a voice note I recorded and send it? No. Campaign voice notes are always generated by ElevenLabs from a script. To sound like yourself, clone your voice in the Voice Library and use that voice. ### Do contacts receive an audio file or a voice message? A voice message, the same kind a person records in WhatsApp. If recording fails, the contact is marked failed instead of getting a file. ### Can each contact hear their own name? Yes. Put `{name}` in the voice script and each contact's clip is generated with their name, from the name field of your list. ### What happens if I run out of ElevenLabs credit mid-campaign? Contacts whose audio cannot be generated are marked failed. If the Delivery Rate Alert is on, the campaign pauses once failures reach its threshold. ### Can follow-ups be voice notes? No. Only a campaign's first message can be a voice note; sequence follow-ups are sent as text or text with an attachment. ### Is my ElevenLabs key the same as my AI key? No. ElevenLabs is set in Settings → Voice Notes; Claude, ChatGPT or DeepSeek keys go in Settings → AI Consultant. Wassuply is independent software. It is not made, endorsed or sponsored by WhatsApp or Meta.