How to Make Your Voice Sound Like Drake, The Weeknd, Taylor Swift and More using AI

Sam Smyers
22 Apr 202314:38

TLDRIn this informative video, Sam Smith explores the technology behind AI-generated singing voices, demonstrating how one can mimic artists like Drake, The Weeknd, and Taylor Swift. Smith discusses the viral song 'Heart on My Sleeve' and introduces tools like Uberduck for text-to-speech and SVC for singing voice conversion. He guides viewers through using Google Collab and a Discord bot to convert vocals, showcasing the process with examples and highlighting the potential applications in the music industry.

Takeaways

  • 😲 The video discusses how AI technology can be used to make one's voice sound like famous artists like Drake, The Weeknd, and Taylor Swift.
  • 🎤 A viral song called 'Heart on My Sleeve' was so convincingly like Drake that Universal Music had to take it down from all platforms.
  • 🔍 There are AI programs like Uberduck for text-to-speech (TTS) and SVC (Singing Voice Conversion) for mimicking singing voices.
  • 🎵 SVC technology can convert one singer's voice to sound like another, provided the original vocal performance is similar to the target artist.
  • 🛠️ A GitHub page hosts the software for singing voice conversion, which can be run using Google Colab without needing extensive technical knowledge.
  • 🎶 The process involves uploading a demo vocal and a voice model of the artist you want to emulate, then running the conversion.
  • 🔗 There are pre-trained voice models available online, or one can train their own model with enough technical understanding.
  • 👤 The video creator used a pre-trained Drake voice model to convert a demo vocal to sound like Drake.
  • 📁 The converted vocal can be downloaded and then edited or mixed into a track using software like Ableton Live.
  • 🤖 There's also a Discord channel with bots that can automate the voice conversion process, though it may be overloaded at times.
  • 🔊 The video creator experimented with different artists' voice models, noting that the quality varies and depends on the training of the model.

Q & A

  • What is the purpose of the video by Sam Smith?

    -The purpose of the video by Sam Smith is to demonstrate how to create a song using AI technology to make one's voice sound like popular artists such as Drake, The Weeknd, and Taylor Swift.

  • What song by Drake and The Weeknd was mentioned in the video that was so convincing it was mistaken for the real artists?

    -The song mentioned in the video is 'Heart on My Sleeve', which was so convincing that Universal Music had to take it down from all platforms because it sounded very much like Drake.

  • What is Uberduck and how is it used in the context of the video?

    -Uberduck is a text-to-speech (TTS) platform where you can type in text and select a voice wrapper to convert the text into a voice. It was mentioned in the video as a tool used by David Guetta to create a performance with an Eminem vocal.

  • What does SVC stand for and what does it do?

    -SVC stands for Singing Voice Conversion. It is a technology that converts one singer's voice to sound like that of another, which is used in the video to imitate artists like Drake and The Weeknd.

  • How does the process of singing voice conversion work as described in the video?

    -The process involves using a program like Soft VC (SVC) where you can input a vocal that sounds similar to the artist you want to imitate. The program then converts the demo singer's voice to match the timbre and style of the target artist.

  • What is Google Colab and how is it used in the video?

    -Google Colab is a machine learning platform that allows users to run programs using Google Drive for storage. In the video, it is used to run the singing voice conversion program without having to download files or learn Python.

  • What is the role of voice models in the singing voice conversion process?

    -Voice models are trained datasets that the singing voice conversion program uses to mimic the voice of a specific artist. They are essential for the program to convert a demo singer's voice to sound like the target artist.

  • How can one obtain a voice model for an artist like Drake?

    -One can obtain a voice model for an artist like Drake by finding a pre-trained model online, as mentioned in the video where someone had already trained a Drake voice model and made it available for others to use.

  • What is the process of using the singing voice conversion program as described in the video?

    -The process involves running the program on Google Colab, selecting a voice model, uploading an audio file of the demo singer, and then running the conversion process. After conversion, the new audio file can be downloaded and used.

  • What is the potential application of singing voice conversion technology in the music industry?

    -The technology can be used for various applications in the music industry, such as creating demo vocals that sound like popular artists for pitch purposes, or for creating unique and innovative sounds by combining different artists' vocal styles.

  • What is the AI bot on Discord mentioned in the video and how does it work?

    -The AI bot on Discord is a tool created by the community that automates the singing voice conversion process. Users can tag the bot with a command and an audio file, and the bot processes the file to convert the voice to the desired artist's style, making it easier for non-technical users to utilize the technology.

Outlines

00:00

🎤 Introduction to AI-Generated Songs

Sam Smith introduces a video exploring the creation of a viral AI-generated song that imitates Drake and The Weeknd. He mentions his other videos and encourages viewers to subscribe and like the video to support the YouTube algorithm. Smith discusses the AI song 'Heart on My Sleeve', which was so convincing that Universal Music had to remove it from platforms. He explains the use of singing voice conversion (SVC) technology, which transforms one singer's voice to sound like another, and mentions the use of Uberduck for text-to-speech conversion. Smith also references David Guetta's use of Eminem's AI-generated vocals at a festival.

05:00

🔧 Demonstrating the Voice Conversion Process

Smith provides a step-by-step guide on how to use singing voice conversion software, specifically mentioning a GitHub page for a program called Soft VC (SVC). He clarifies that he is not a technical expert and apologizes for any potential inaccuracies. Smith describes using Google Collab to run the program without needing to download files or learn Python. He shares his process of finding a splice vocal similar to Drake's and demonstrates how to input the vocal into the program called Sovitz. He also discusses the possibility of training one's own voice model using a dataset of a particular artist's vocals.

10:02

🎶 Testing AI Voice Conversion with Different Artists

Smith shares his experiences with testing the AI voice conversion software using different voice models. He successfully converts a demo vocal to sound like Drake and then imports it into Ableton Live for mixing. He notes some imperfections and glitches in the converted vocals and suggests that the quality depends on the training model. Smith also mentions an easier method involving a Discord channel with bots that automate the conversion process. He tests various artists' models, including Taylor Swift and Ariana Grande, and comments on the quality of the results. He concludes by discussing the potential applications of SVC technology in the music industry, such as pitch-perfecting demos for pop stars.

Mindmap

Keywords

💡AI

AI stands for Artificial Intelligence, which refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the context of the video, AI is used to create music that sounds like it is performed by famous artists such as Drake and The Weeknd. The script mentions the use of AI in creating a viral song that was so convincing it had to be taken down by Universal Music.

💡Drake

Drake is a popular Canadian rapper, singer, and songwriter known for his distinctive voice and style. In the video script, Drake's voice is mentioned as an example of how AI can be used to make another person's voice sound like his. The script discusses a song called 'Heart on My Sleeve' that was created using AI to imitate Drake's voice convincingly.

💡The Weeknd

The Weeknd is a Canadian singer, songwriter, and record producer known for his unique vocal style and R&B influenced music. Similar to Drake, The Weeknd's voice is referenced in the script as another example of an artist whose voice was convincingly replicated using AI technology.

💡TTS

TTS stands for Text-to-Speech, a technology that converts written text into audible speech. The script mentions Uberduck, a TTS service that allows users to type in text and select a voice wrapper to convert the text into speech, imitating artists like Eminem for rapping.

💡SVC

SVC stands for Singing Voice Conversion, a technology that converts one singer's voice to sound like another. The script explains that SVC was used to create the 'Heart on My Sleeve' song, making it sound as if it was sung by Drake and The Weeknd. This technology is central to the video's theme of AI in music production.

💡Uberduck

Uberduck is a website mentioned in the script that provides TTS services. It allows users to input text and select a voice model to generate speech. The script uses Uberduck as an example of how AI can be used to imitate voices of famous artists like Eminem.

💡GitHub

GitHub is a platform for software development that provides a version control system and collaboration tools. In the script, GitHub is mentioned as the source for a popular singing voice conversion program called 'SVC now', which can be run using Google Colab.

💡Google Colab

Google Colab is a cloud-based development environment designed for machine learning. The script describes how Google Colab can be used to run the 'SVC now' program, which facilitates the process of singing voice conversion without requiring users to download files or learn Python.

💡Acapella

An acapella is a version of a song that features only the vocal track without any instrumental accompaniment. The script suggests that to create a custom voice model for SVC, one could use a series of acapellas from an artist like Billie Eilish, run them through a training model to create a dataset.

💡Discord

Discord is a VoIP and instant messaging platform that is popular among gamers but also used for various communities. In the script, a Discord channel is mentioned where bots have been set up to facilitate the singing voice conversion process, making it even more accessible for users.

💡Timbre Transfer

Timbre Transfer refers to the process of changing the quality or color of a sound, often used in music production to emulate the sound of different instruments or voices. The script discusses how this concept is applied in the context of AI, allowing one person's voice to be converted to have the timbre of another, such as making a demo singer's voice sound like Taylor Swift's.

Highlights

Sam Smith demonstrates how to create AI-generated songs that mimic the voices of famous artists like Drake and The Weeknd.

The viral song 'Heart on My Sleeve' was so convincing that Universal Music had to take it down from all platforms.

Different versions of Drake AI covers have emerged, showcasing the potential of AI in music.

Uberduck is a text-to-speech platform that allows users to type in text and convert it into a voice of their choice.

Singing Voice Conversion (SVC) technology can convert one singer's voice to sound like another.

SVC requires a vocal that sounds similar to the artist for more convincing results.

Soft VCVITS is a popular program for singing voice conversion, available on GitHub.

Google Colab can be used to run the Soft VCVITS program without needing to download files or learn Python.

A splice vocal that sounds similar to Drake is used to demonstrate the voice conversion process.

Voice models can be inputted into the program to convert a demo singer's voice to sound like a specific artist.

There are pre-trained voice models available online, or users can train their own models with enough technical knowledge.

Discord channels with AI bots have been set up to simplify the voice conversion process for users.

The AI bots can process audio files and convert them to sound like specific artists with a simple command.

The quality of the AI-generated voice can vary depending on the training and data input into the model.

Tonal transfer or timbre transfer is a concept used in voice conversion, allowing one person's voice to be transformed into another's.

AI voice conversion has practical applications in the music industry, such as pitch-perfecting demos for artists.

Sam Smith encourages developers to make the voice conversion process more accessible and user-friendly.

The video concludes with a demonstration of the AI-generated voice conversion applied to various artists' styles.