使用条款

These terms cover your use of audiofilter.net and the six tools on it. Using the site means you accept them. They are written to be read, not to be skipped — there is not much to them, because there is no account, no payment and no data of yours on our side.

1. What the service is

Audiofilter provides six free browser-based audio tools: a vocal remover, a noise reducer, a text-to-speech generator, an audio converter and MP3 compressor, a voice changer, and AI Speech to Text. All six run inside your browser on your own device. No account is required and nothing is charged. See Pricing for what free means in practice and what the limits are.

2. The files you process, and who is responsible for them

You are. The tools do not check what you feed them and could not — your file never reaches us. You are responsible for having the right to process whatever you put in.

Because processing happens entirely on your machine, we have no ability to review, moderate, take down or restore anything. There is nothing on our side to act on.

3. Acceptable use

Do not use the site to:

• Infringe anyone's copyright, moral rights, trade marks, or rights of publicity.
• Clone a person's voice to impersonate them, to deceive, or to create material attributed to them that they did not say — see voice conversion and voice cloning below.
• Produce material that is unlawful where you are.
• Attack the site — automated scraping at a rate that degrades it for others, attempts to circumvent hosting protections, or probing for vulnerabilities without contacting us first.

We cannot enforce the first three, and we say so plainly rather than pretending to a moderation capability the architecture does not allow. They remain your obligation.

Voice conversion and voice cloning

Two tools here can make audio that sounds like a real person saying something they never said: the voice changer rebuilds one recording in another person's voice, and the text-to-speech tool can clone a voice from a short sample. That is the use of this site most capable of hurting someone who is not in the room, so it gets rules of its own, on top of the list above.

• Have the voice owner's consent. Use your own voice, one of the built-in voices, or the voice of someone who has agreed to it being reproduced this way. Being allowed to have a recording is not the same as being allowed to reproduce the voice in it.
• Have the speaker's consent too. The voice changer keeps the words, timing and delivery of the source recording and changes only who seems to be saying them. If the source is someone else speaking, their performance is in the result, and they have a say in that.
• No fraud, and no getting past voice checks. Do not use converted or cloned audio to pass a voice-verification or voice-biometric check — phone banking, account recovery, access control — or to obtain money, access or information by posing as someone else.
• No harassment, defamation or intimate content. Do not use a real person's voice to threaten, harass or humiliate them, to put false statements in their mouth, or to make sexual or intimate content of them without their consent.
• Say it is synthetic when you publish it. If you share converted or cloned audio that could be taken for a real person, label it as generated. In some jurisdictions — the European Union among them — that disclosure can be a legal requirement rather than a courtesy.

As with the rest of this section, none of this can be checked from our side: the audio is processed in your browser and never reaches us. That is a description of the architecture, not a permission. Where a use sits in a grey area and you are unsure whether it is lawful where you are, do not do it until you have asked someone qualified to tell you.

4. No warranty

The tools are provided as is, without warranty of any kind, express or implied, including fitness for a particular purpose.

Concretely, and not as boilerplate:

• Results vary by input. Machine separation is imperfect. A dense mix, a heavily processed vocal or a low-bitrate source can produce audible artefacts, and no setting fixes that.
• It depends on your device. Processing uses your CPU and memory. An older machine may be slow; a device short on memory may fail outright. The text-to-speech tool additionally requires a browser feature (SharedArrayBuffer) that some browsers and some embedded webviews do not provide.
• Output is not archival. Download and keep anything you want to keep. Nothing is stored for you, so a closed tab is a lost result.
• The site may change or stop. Tools, limits and models may be altered or withdrawn without notice.

5. Limitation of liability

To the fullest extent the law where you are allows, Audiofilter is not liable for indirect, incidental, special or consequential loss arising from your use of the site — including lost work, lost time, lost data, or any claim brought against you over material you produced with it.

Nothing here limits liability that cannot lawfully be limited.

6. Third-party software

The tools include open-source components, distributed under permissive licences including the MIT Licence and the Apache License 2.0. Those licences permit this use and require their notices be retained; copies of the applicable licence texts and notices are available on request at the contact address below. Those components are provided by their authors without warranty, and their licences do not imply any endorsement of this site by them.

7. Intellectual property in the site itself

The site's name, design, written content and diagrams are ours. The output you produce with the tools is not — we claim no rights over your results and grant ourselves no licence to them, which would be difficult in any case given we never see them.

8. Changes to these terms

These terms may be updated. The date at the top of this page shows when they last were. Continuing to use the site after a change means accepting the revised terms.

9. Contact

Questions about these terms, licence notices, or a security issue you have found: audiofilternet@outlook.com.