Free · local processing · no account
Build your own audio workflow
Connect noise reduction, trimming, equalization, volume and audio mixing in the order you need. Choose the inputs for each module, listen to the results and download your finished audio.
Designed for a PC or a large screen. Connecting and configuring modules is easier with a mouse; please use a computer for the best experience.
Clearing processing results keeps originals and undo history. Releasing history clears undo steps but keeps current audio and results. Temporary audio stays in this browser, separately from AI model caches.
Add an audio source or start with a template.
Audio pipeline
Follow the arrows from your audio sources to the final output. Branches come together at the mixer.
Listen and compare stages
Click a module or choose a preview stage to inspect its result. Large results are stored temporarily in this browser and loaded when needed. Refreshing still requires reopening source audio.
Downloads use the module connected to Output audio in the pipeline. Selecting a preview stage does not change the exported audio.
An AI music filter and cutter in one workflow
Start with vocal separation to keep vocals or accompaniment, connect Trim audio to keep or remove a section, then use EQ to shape the sound. AI modules handle separation or noise reduction; trimming and effects use local audio processing. Select any completed stage to check its result before downloading.
Vocal remover · Audio Cutter · Audio Equalizer
How to build an audio workflow
Place modules in empty canvas space first, then drag their ports to connect them. New modules remain unconnected until you choose their inputs. Each module connects to only one downstream module. Choose any of its four sides for the output. Processing modules accept one input; each mixer input point accepts multiple connections. Drag a line segment to adjust its route. Use Data flow demo to see how connected modules pass audio to the output. Switch between curves and straight lines with right-angle bends. You can also click an output port followed by an input port. Select a module to edit its settings; loops are rejected.
Add source modules and choose an audio file for each one. Add processing modules and connect them in the pipeline. Module settings display the connected inputs; change connections in the diagram. A mix takes at least two inputs and lets you adjust source segments, volume and timeline positions. Connect the final module to Output audio and run the flow.
Update only the steps that changed
When you change a setting, the affected module and its downstream results need updating. Completed upstream results are reused. Use Skip this module to compare a single-input module with its input. Cancel stops active processing and retains completed stages.
Save settings and keep your audio local
Save flow adds your workflow to the top of the template list and downloads a JSON file. Your current modules, connections, settings and layout are remembered in this browser. Open flow file also adds an imported flow to your templates. Flow files and saved templates do not contain audio. Large audio results may be stored temporarily in this browser. Unused data is removed automatically; expired data from interrupted sessions is cleaned on a later visit. Reopen source audio after refreshing or switching templates. Undo last action restores recent edits, including Restore initial state.
Processing limits and audio quality
Noise reduction and vocal separation accept up to 300 seconds and 300 MB per input. Speech recognition accepts up to 300 seconds. Voice changing accepts up to 10 seconds and 60 MB, plus a reference of up to 30 seconds and 60 MB. Speech synthesis accepts up to 500 characters; custom reference voices must be 3–15 seconds and no larger than 25 MB. Limits are checked after trimming, mixing and speed changes, then checked again against the actual input when running. Intermediate audio sizes are measured as uncompressed PCM. Other tools depend on available device memory.
All site tools are available in the searchable module library. Speech to text outputs editable text; connect it to Text to speech to generate English audio, or download the transcript. Voice changing requires a separate reference recording. Text entered in module settings is included when you save a flow; audio and reference recordings must be reopened.