AudioRinse

AudioRinse fixes recordings you made: an interview with an air conditioner behind it, a lecture in an echoey room, a voice memo, a podcast episode or the sound of a video. Claude can't hear audio; this connector checks it and cleans it.

Tools:

• analyze_audio checks a recording and says what's wrong in plain words: background noise, room echo, loudness against podcast and YouTube targets, possible clipping, and whether it's speech. It recommends a fix, or says cleaning won't help. Checks are free.
• clean_audio removes background noise, reduces room echo and reverb, or clears up a voice and levels it for podcasts or YouTube. It returns a download link and a page that switches between the original and the cleaned version. Videos keep their picture.
• upload_audio returns a private upload link for a file on your device, because files attached to a chat can't be passed to connectors.
• check_job returns the result for a longer file that is still being cleaned.
• my_usage and whoami show the signed-in account, its plan and what's left this month.

The cleaning is filtering, not generation: the result is your own recording with noise and echo turned down. No speech is synthesised, changed or cloned. It's built for speech, not music.

It uses your AudioRinse account, created the first time you sign in (email or Google). Free: 3 cleanings a month, shared with downloads on audiorinse.com, for audio files up to 10 minutes; checks are always free. Pro ($7.99 a month or $39.99 a year): 20 hours a month in Claude, files up to 40 minutes, video and WAV, plus unlimited downloads on the website. Details at audiorinse.com/pricing.

Privacy: files are cleaned on our own servers by an open-source model and are never sent to another AI service. Uploads are deleted as soon as they're cleaned, results after 24 hours. Recordings are never listened to or used for training. The connector doesn't read your conversations, memory or other files.