ElevenLabs Noise Remover vs Sound Effects v2
Choose Noise Remover when speech already exists inside a noisy MP3 or WAV and you need an isolated audio output. Choose Sound Effects v2 when no source recording exists and you need to generate a new non-speech event, ambience, impact, or loop. Both use ElevenLabs through fal and the same private audio output path, but they do opposite jobs.
ElevenLabs Noise Remover
Isolates voice from one existing private audio file through fal's ElevenLabs Audio Isolation endpoint.
- Best fit
- Interviews, voice notes, dialogue, or spoken recordings with unwanted background sound.
- Output
- One privately stored isolated audio file
- Cost
- About $0.10 per verified input minute; 1-500 credits
ElevenLabs Sound Effects v2
Creates one new private MP3 sound effect from a non-speech text description.
- Best fit
- Impacts, interfaces, transitions, foley, ambience, textures, and seamless loops.
- Output
- One privately stored MP3 at 44.1 kHz and 128 kbps
- Cost
- About $0.002 per output second; 1-4 credits
Compare the same decision criteria
The decision starts from whether useful speech already exists in a source file. The two routes never substitute for each other or share a credit hold.
| Criterion | ElevenLabs Noise Remover | ElevenLabs Sound Effects v2 | Decision rule |
|---|---|---|---|
| Primary job | Remove background sound around existing speech | Generate a new non-speech sound | Choose from whether a source recording already exists. |
| Input | Private MP3 or WAV | Text description and output controls | Noise Remover never accepts a text-only cleanup request. |
| Duration | Verified source length from 1 second to 60 minutes | Requested whole-second output from 1 to 22 seconds | Review the correct duration before comparing price. |
| Pricing unit | About $0.10 per input minute, prorated by verified second | About $0.002 per output second | Input duration and requested output duration are different supplier meters. |
| Credits | 1-500 | 1-4 | A one-minute cleanup costs 9 credits; a five-second sound effect costs 1. |
| Endpoint | fal-ai/elevenlabs/audio-isolation | fal-ai/elevenlabs/sound-effects/v2 | No Music, cloning, extra TTS, eleven-v3, or deprecated Sound Effects v1 route is bundled. |
On smaller screens, this matrix scrolls inside its own frame. The page itself does not scroll sideways.
Reserved audio proof without borrowed cleanup or sound output
Both proof slots remain reserved until Kanvora approves source-tracked output from the exact endpoint. An isolated voice track is never shown as a generated sound effect or vice versa.
Noise Remover source plan
No text prompt is sent. Upload one private MP3 or WAV and review the verified duration before submission.
A verified 60-second source costs 9 credits. The real file stays behind the authenticated same-origin upload boundary.
Open this directionSound Effects v2 direction
Heavy wooden door closing in a large stone hall, sharp impact, short low-frequency tail, no speech and no music.
Five seconds, balanced influence, and one-shot output cost 1 credit.
Open this directionThe source plan and sound direction do not prove cleanup quality, event accuracy, artifact removal, loop quality, loudness, or a universal winner.
Choose from the audio you have and the audio you need
These scenarios keep cleanup and generation separate without quietly adding AI Music, another TTS model, cloning, or eleven-v3.
A 60-second interview has steady fan noise behind clear speech.
Signal: The spoken content already exists and must remain.
Choose Noise Remover.Open the cleanup workbenchThe route accepts the private source and charges 9 credits from its verified one-minute duration.
A game menu needs a compact confirmation tone.
Signal: There is no source recording to clean.
Choose Sound Effects v2.Create the sound directionA text description, short duration, and one-shot ending match the generation endpoint.
A song needs the lead vocal separated from the instrumental.
Signal: Music-source separation is the primary requirement.
Do not treat Noise Remover as a guaranteed stem separator.Review the cleanup limitsElevenLabs documents voice isolation and warns that music vocals are not its optimized use case.
A project needs a new full song.
Signal: Composition, lyrics, and music generation are required.
Use neither route.Review live audio toolsAI Music is not shipping in this catalog round.
Switching audio jobs creates a separate paid request
Wait for active or uncertain jobs to settle. Changing between cleanup and generation changes input contract, endpoint, supplier meter, and credit hold.
Switching and handoff costs
- Noise Remover bills verified input duration at about $0.10 per minute, prorated by second.
- Sound Effects v2 bills requested output duration at about $0.002 per second.
- Kanvora rounds each supplier total up to a positive whole-credit hold under the same $0.012 coverage formula.
- Neither request triggers Music, extra TTS, cloning, eleven-v3, dialogue, or the deprecated Sound Effects v1 endpoint.
What the comparison cannot prove
- Noise Remover does not guarantee studio-clean speech or complete background removal.
- Isolation can leave artifacts or remove wanted voice content.
- Sound Effects v2 does not guarantee a specific recording, acoustic space, or loop seam.
- Reserved proof slots are not quality benchmarks.
Clean existing speech with Noise Remover. Create new sound with Sound Effects.
Both CTAs carry one legal live operation into the authenticated workbench. Private upload, exact hold, and stored audio output stay at the real request boundary.