zsxkib / realistic-voice-cloning

Create song covers with any RVC v2 trained AI voice from audio files.

Cold

Public
634.7K runs
T4
GitHub
License

Run with an API

Playground API Examples README Versions

Input

Video Player is loading.

Current Time 00:00:000

Duration 00:00:000

Loaded: 0%

Stream Type LIVE

Remaining Time 00:00:000

song_input

file

Upload your audio file here.

rvc_model

string

RVC model for a specific voice. If using a custom model, this should match the name of the downloaded model. If a 'custom_rvc_model_download_url' is provided, this will be automatically set to the name of the downloaded model.

Default: "Squidward"

custom_rvc_model_download_url

string

Shift + Return to add a new line

URL to download a custom RVC model. If provided, the model will be downloaded (if it doesn't already exist) and used for prediction, regardless of the 'rvc_model' value.

pitch_change

string

Adjust pitch of AI vocals. Options: `no-change`, `male-to-female`, `female-to-male`.

Default: "no-change"

index_rate

number

(minimum: 0, maximum: 1)

Control how much of the AI's accent to leave in the vocals.

Default: 0.5

filter_radius

integer

(minimum: 0, maximum: 7)

If >=3: apply median filtering median filtering to the harvested pitch results.

Default: 3

rms_mix_rate

number

(minimum: 0, maximum: 1)

Control how much to use the original vocal's loudness (0) or a fixed loudness (1).

Default: 0.25

pitch_detection_algorithm

string

Best option is rmvpe (clarity in vocals), then mangio-crepe (smoother vocals).

Default: "rmvpe"

crepe_hop_length

integer

When `pitch_detection_algo` is set to `mangio-crepe`, this controls how often it checks for pitch changes in milliseconds. Lower values lead to longer conversions and higher risk of voice cracks, but better pitch accuracy.

Default: 128

protect

number

(minimum: 0, maximum: 0.5)

Control how much of the original vocals' breath and voiceless consonants to leave in the AI vocals. Set 0.5 to disable.

Default: 0.33

main_vocals_volume_change

number

Control volume of main AI vocals. Use -3 to decrease the volume by 3 decibels, or 3 to increase the volume by 3 decibels.

Default: 0

backup_vocals_volume_change

number

Control volume of backup AI vocals.

Default: 0

instrumental_volume_change

number

Control volume of the background music/instrumentals.

Default: 0

pitch_change_all

number

Change pitch/key of background music, backup vocals and AI vocals in semitones. Reduces sound quality slightly.

Default: 0

reverb_size

number

(minimum: 0, maximum: 1)

The larger the room, the longer the reverb time.

Default: 0.15

reverb_wetness

number

(minimum: 0, maximum: 1)

Level of AI vocals with reverb.

Default: 0.2

reverb_dryness

number

(minimum: 0, maximum: 1)

Level of AI vocals without reverb.

Default: 0.8

reverb_damping

number

(minimum: 0, maximum: 1)

Absorption of high frequencies in the reverb.

Default: 0.7

output_format

string

wav for best quality and large file size, mp3 for decent quality and small file size.

Default: "mp3"

Run this model in Node.js with one line of code:

npx create-replicate --model=zsxkib/realistic-voice-cloning

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run zsxkib/realistic-voice-cloning using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "zsxkib/realistic-voice-cloning:0a9c7c558af4c0f20667c1bd1260ce32a2879944a0b9e44e1398660c077b1550",
  {
    input: {
      protect: 0.33,
      rvc_model: "Squidward",
      index_rate: 0.5,
      song_input: "https://replicate.delivery/pbxt/JsPIizFfRy54Jk5LuXdnrNdV1JHJ6oLmPPdRuIfh3lvpoNai/gangnam.mp3",
      reverb_size: 0.15,
      pitch_change: "no-change",
      rms_mix_rate: 0.25,
      filter_radius: 3,
      output_format: "mp3",
      reverb_damping: 0.7,
      reverb_dryness: 0.8,
      reverb_wetness: 0.2,
      crepe_hop_length: 128,
      pitch_change_all: 0,
      main_vocals_volume_change: 10,
      pitch_detection_algorithm: "rmvpe",
      instrumental_volume_change: 0,
      backup_vocals_volume_change: 0
    }
  }
);
console.log(output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run zsxkib/realistic-voice-cloning using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "zsxkib/realistic-voice-cloning:0a9c7c558af4c0f20667c1bd1260ce32a2879944a0b9e44e1398660c077b1550",
    input={
        "protect": 0.33,
        "rvc_model": "Squidward",
        "index_rate": 0.5,
        "song_input": "https://replicate.delivery/pbxt/JsPIizFfRy54Jk5LuXdnrNdV1JHJ6oLmPPdRuIfh3lvpoNai/gangnam.mp3",
        "reverb_size": 0.15,
        "pitch_change": "no-change",
        "rms_mix_rate": 0.25,
        "filter_radius": 3,
        "output_format": "mp3",
        "reverb_damping": 0.7,
        "reverb_dryness": 0.8,
        "reverb_wetness": 0.2,
        "crepe_hop_length": 128,
        "pitch_change_all": 0,
        "main_vocals_volume_change": 10,
        "pitch_detection_algorithm": "rmvpe",
        "instrumental_volume_change": 0,
        "backup_vocals_volume_change": 0
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run zsxkib/realistic-voice-cloning using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "0a9c7c558af4c0f20667c1bd1260ce32a2879944a0b9e44e1398660c077b1550",
    "input": {
      "protect": 0.33,
      "rvc_model": "Squidward",
      "index_rate": 0.5,
      "song_input": "https://replicate.delivery/pbxt/JsPIizFfRy54Jk5LuXdnrNdV1JHJ6oLmPPdRuIfh3lvpoNai/gangnam.mp3",
      "reverb_size": 0.15,
      "pitch_change": "no-change",
      "rms_mix_rate": 0.25,
      "filter_radius": 3,
      "output_format": "mp3",
      "reverb_damping": 0.7,
      "reverb_dryness": 0.8,
      "reverb_wetness": 0.2,
      "crepe_hop_length": 128,
      "pitch_change_all": 0,
      "main_vocals_volume_change": 10,
      "pitch_detection_algorithm": "rmvpe",
      "instrumental_volume_change": 0,
      "backup_vocals_volume_change": 0
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

You can run this model locally using Cog. First, install Cog:

brew install cog

If you don’t have Homebrew, there are other installation options available.

Run this to download the model and run it in your local environment:

cog predict r8.im/zsxkib/realistic-voice-cloning@sha256:0a9c7c558af4c0f20667c1bd1260ce32a2879944a0b9e44e1398660c077b1550 \
  -i 'protect=0.33' \
  -i 'rvc_model="Squidward"' \
  -i 'index_rate=0.5' \
  -i 'song_input="https://replicate.delivery/pbxt/JsPIizFfRy54Jk5LuXdnrNdV1JHJ6oLmPPdRuIfh3lvpoNai/gangnam.mp3"' \
  -i 'reverb_size=0.15' \
  -i 'pitch_change="no-change"' \
  -i 'rms_mix_rate=0.25' \
  -i 'filter_radius=3' \
  -i 'output_format="mp3"' \
  -i 'reverb_damping=0.7' \
  -i 'reverb_dryness=0.8' \
  -i 'reverb_wetness=0.2' \
  -i 'crepe_hop_length=128' \
  -i 'pitch_change_all=0' \
  -i 'main_vocals_volume_change=10' \
  -i 'pitch_detection_algorithm="rmvpe"' \
  -i 'instrumental_volume_change=0' \
  -i 'backup_vocals_volume_change=0'

To learn more, take a look at the Cog documentation.

Run this to download the model and run it in your local environment:

docker run -d -p 5000:5000 --gpus=all r8.im/zsxkib/realistic-voice-cloning@sha256:0a9c7c558af4c0f20667c1bd1260ce32a2879944a0b9e44e1398660c077b1550
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "protect": 0.33,
      "rvc_model": "Squidward",
      "index_rate": 0.5,
      "song_input": "https://replicate.delivery/pbxt/JsPIizFfRy54Jk5LuXdnrNdV1JHJ6oLmPPdRuIfh3lvpoNai/gangnam.mp3",
      "reverb_size": 0.15,
      "pitch_change": "no-change",
      "rms_mix_rate": 0.25,
      "filter_radius": 3,
      "output_format": "mp3",
      "reverb_damping": 0.7,
      "reverb_dryness": 0.8,
      "reverb_wetness": 0.2,
      "crepe_hop_length": 128,
      "pitch_change_all": 0,
      "main_vocals_volume_change": 10,
      "pitch_detection_algorithm": "rmvpe",
      "instrumental_volume_change": 0,
      "backup_vocals_volume_change": 0
    }
  }' \
  http://localhost:5000/predictions

To learn more, take a look at the Cog documentation.

Output

Video Player is loading.

Current Time 00:00:000

Duration 00:00:000

Loaded: 0%

Stream Type LIVE

Remaining Time 00:00:000

Generated in

2.8 seconds

Tweak itReport

This example was created by a different version, zsxkib/realistic-voice-cloning:a0076ea1.

Run time and cost

This model costs approximately $0.044 to run on Replicate, or 22 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia T4 GPU hardware. Predictions typically complete within 4 minutes. The predict time for this model varies significantly based on the inputs.

Readme

Realistic Voice Cloning v2 (RVC v2) 🎙️🎭

Overview 🔊

Realistic Voice Cloning v2 (RVC v2) is a voice-to-voice model that transforms an input voice into a target voice. This tool is built upon the amazing work of SociallyIneptWeeb. We’ve wrapped his AICoverGen repo to work on Replicate! Allowing us to create AI songs with any RVCv2 model on the web!

Support SociallyIneptWeeb and learn more about his work through:

Pre-loaded Model 🦑

The RVC v2 comes pre-loaded with the Squidward voice model for immediate use.

Getting Other Voices 💥

You can expand your RVC v2’s capabilities by downloading additional models from places like QuickWick’s Music-AI-Voices Hugging Face Repository. Here’s how:

Visit the repository and go to Files and Versions
Select a model, such as the Music-AI-Voices/2Pac Tupac (RVC) 150 Epoch. It’s important to pick a model which has “(RVC)” in the name
Then click on the .zip file to access the model’s download page, e.g., 2Pac Tupac (RVC) 150 Epoch
Right click on the download button and copy the link address - e.g. download URL: https://huggingface.co/QuickWick/Music-AI-Voices/resolve/main/2Pac%20Tupac%20(RVC)%20150%20Epoch/2Pac%20Tupac%20(RVC)%20150%20Epoch.zip?download=true as custom_rvc_model_download_url
Select the model to use (once it’s automatically downloaded) rvc_model = "CUSTOM"

Note: Ensure that the rvc_model and custom_rvc_model_download_name match, as they represent the chosen model and the name of the downloaded model file, respectively.

Another good websites to find RVC (v2) models are voice-models

Terms of Use 📚

The use of the converted voice for the following purposes is prohibited.

Criticizing or attacking individuals.
Advocating for or opposing specific political positions, religions, or ideologies.
Publicly displaying strongly stimulating expressions without proper zoning.
Selling of voice models and generated voice clips.
Impersonation of the original owner of the voice with malicious intentions to harm/hurt others.
Fraudulent purposes that lead to identity theft or fraudulent phone calls.

Disclaimer ‼️

I am not liable for any direct, indirect, consequential, incidental, or special damages arising out of or in any way connected with the use/misuse or inability to use this software.