WhatsApp Voice Cloning n8n Workflow: Fish Audio + FreeConvert

Clone WhatsApp Voice Notes and Reply Back with Fish Audio (n8n Workflow + FreeConvert)

Turn incoming WhatsApp voice notes into an actual Fish Audio private TTS voice model—then use it to generate and send voice replies on demand when users send /tts text commands. This n8n workflow automates the full pipeline: WhatsApp voice note → FreeConvert MP3 → Fish Audio voice cloning → Google Sheets logging → WhatsApp audio responses.

What this workflow does

• Receives WhatsApp messages via an n8n WhatsApp trigger, extracting sender phone number, message type, media ID, and any text body.
• For voice notes: fetches the media URL using the Facebook Graph API, downloads the audio, and uploads it to FreeConvert for conversion to MP3.
• Waits for conversion completion by polling the FreeConvert job status until it reports completed.
• Creates a Fish Audio voice clone by uploading the MP3 audio to Fish Audio to generate a private TTS voice model.
• Logs model details by upserting the returned Fish Audio model ID and name into Google Sheets, then sends a WhatsApp confirmation message back to the sender.
• For text commands: when the incoming text contains /tts, it runs a sub-workflow to generate speech with Fish Audio, sends the resulting audio back on WhatsApp.

Use cases

• Build an AI voice assistant where customers provide a short WhatsApp voice note and then receive voice replies generated from their own voice.
• Create an onboarding flow: users send a voice sample on WhatsApp, you clone it once, and confirm readiness in the same chat.
• Automation engineering demo project for end-to-end WhatsApp + FreeConvert + Fish Audio + Google Sheets integrations in n8n.

Technical details

• WhatsApp Cloud API trigger for new messages and sending voice replies.
• Facebook Graph API to retrieve voice note media URLs.
• FreeConvert (HTTP Bearer token) to convert audio to MP3 and poll job status.
• Fish Audio to train a private TTS voice model from the converted MP3.
• Google Sheets upsert to store Fish Audio model ID/name.
• Workflow logic uses nodes such as if, set, code, wait, plus WhatsApp-related steps (with sticky notes for guidance).