Removing Panned Vocals: Techniques For A Cleaner Mix

Removing vocals from a song is a common task for music producers and audio engineers, and it can be done using various software tools and techniques. While it is challenging to completely remove vocals without affecting the overall sound quality, several methods and tools can help achieve satisfactory results. This includes using phase cancellation technology, EQ plugins, and advanced audio editing software with vocal reduction and isolation features. Additionally, AI-powered tools and online services have made it easier to separate vocals from instrumentals, creating karaoke versions or remixes.

Using AI to separate vocals

AI technology has revolutionized the way we create and consume music, and one of its most impressive applications is in vocal removal and isolation. With AI algorithms, you can now easily separate vocals from instrumentals, creating karaoke versions or instrumental tracks of your favorite songs. This process is known as music source separation, and it's an incredibly useful technique for musicians, DJs, and music enthusiasts.

One of the most popular AI tools for vocal removal is LALAL.AI. This web-based application allows you to upload any song and choose from various stem separation types, including vocal and instrumental, voice and noise, or even individual instruments like drums, bass, piano, and guitar. LALAL.AI uses advanced machine learning algorithms to process audio signals and separate the vocals from the instrumentals effectively. The best part? It offers a free plan so you can try out the service before upgrading to process more files and get faster results.

Another notable AI tool for vocal separation is Moises. With its vocal remover feature, you can effortlessly separate vocals, guitar, drums, bass, and other instruments from any song. Moises offers a Pro Plan that includes Hi-Fi Audio Separation, giving you pristine results for your music production projects. The Moises app is available on iOS, Android, and desktop, making it accessible to a wide range of users.

For those who prefer a desktop application, Audacity is a free, open-source digital audio workstation (DAW) that includes vocal reduction and isolation tools. Audacity's vocal removal effect is based on phase cancellation, where the left and right audio channels are inverted to cancel out the vocals, which are typically placed in the center of the stereo field. While this method may not always produce perfect results, it's still a valuable tool for beginners and professionals looking for a free option.

In addition to these tools, there are also online services like Vocal Remover and Isolation, which use AI to create a karaoke version and an a cappella version of your chosen song. This service is completely free, and processing usually takes about 10 seconds.

With these AI tools, the process of separating vocals has become more accessible and efficient than ever before. Whether you're a musician looking to create instrumental versions of your songs or a music enthusiast wanting to sing along with your favorite tracks, AI vocal removal and isolation technology has got you covered.

Manually separating the stereo field

To do this, first, open the audio file in Audacity. Then, select the area of the audio file where you want to remove the vocals. Next, split the stereo track into two mono tracks by selecting "Split Stereo Track" from the track's dropdown menu. This will result in two identical mono tracks.

Now, invert one of the mono tracks by selecting it and choosing "Effect" -> "Invert". This will create a mirror image of the audio wave, causing the two waves to cancel each other out through phase cancellation.

Finally, play back the result. This process will remove everything panned in the centre, including vocals, and return a dual mono result. However, keep in mind that this method may not completely remove the vocals, especially if there are backing vocals or reverb present, as these can spread sound sources and make them harder to extract.

Using EQ plugins

EQ plugins are a powerful tool for shaping the sound of vocals in a mix. While there is no one-size-fits-all approach due to the uniqueness of each voice and recording, there are some general techniques that can help you remove panned vocals using EQ plugins. Here are some detailed instructions on how to do this:

Understanding the Frequency Spectrum

Before adjusting any EQ settings, it's important to understand the frequency spectrum and how it relates to the human voice. The audible frequency spectrum for humans ranges from 20 Hz to 20 kHz, and different frequency ranges contribute to the overall character of a vocal performance. For example, the lowest frequencies of a vocal typically fall between 100 Hz and 400 Hz and provide the "body" and "weight" of the vocal. On the other hand, the high frequencies above 5 kHz add clarity and allow the vocal to cut through other instruments in the mix.

Using EQ to Remove Panned Vocals

To remove panned vocals using EQ plugins, follow these steps:

• Identify the Vocal Range: Determine the vocal range of the singer, which can fall into categories such as bass, baritone, tenor, alto, mezzo-soprano, or soprano. This will help you identify the fundamental frequency range you'll be working with.
• Isolate the Vocals: Use a utility plugin to isolate the center of the mix, where vocals are typically panned. This will allow you to focus solely on the vocal frequencies.
• Remove Unwanted Low End: Engage a high-pass filter (HPF) to remove low frequencies below 100 Hz. This range typically contains room rumble, microphone handling noise, and other unwanted sounds that can muddy up the mix.
• Balance Body and Warmth: The frequency range between 100 Hz and 400 Hz is crucial for the body and warmth of the vocal. Use a bell filter with a medium to tight Q to identify and cut any overly resonant or troublesome frequencies in this range.
• Remove Hollow or Boxy Frequencies: The range between 400 Hz and 800 Hz contributes to the character of the vocal. Use the "boost and sweep" method to identify and cut any unwanted "papery" or boxy resonances that can make the vocal sound unnatural.
• Address the Mid Range: The mid-range frequencies between 800 Hz and 1.5 kHz can accentuate the throatiness of the vocal, making it sound honky or nasally. Use the "boost and sweep" method to identify any unwanted resonances and cut them with a tighter Q to avoid affecting nearby frequencies.
• Improve Presence and Intelligibility: The range between 1.5 kHz and 5 kHz is crucial for the clarity and intelligibility of the vocal. Small adjustments of 1-2 dB are recommended in this range, as too much can make the vocal harsh, while too little can make it sound flat or muffled.
• Address Sibilance: Sibilance, the bright resonance of "S" and "T" sounds, typically falls between 5 kHz and 8 kHz. While boosting this range can increase sibilance, cutting it can make the vocal sound dull and lifeless. Consider using a de-esser to address sibilance without affecting the overall vocal performance.
• Add Sparkle: To add brilliance and sparkle to the vocal, boost frequencies between 8 kHz and 12 kHz. This can help the vocal feel more open and exciting without affecting the overall tone.
• Fine-tune with a Second EQ: Consider using a different EQ plugin, such as a vintage analog EQ, to add character to the high frequencies above 12 kHz. Boost these frequencies slightly to enhance the airiness of the vocal without making it sound harsh.

Remember, these are general guidelines, and each vocal performance is unique. Always use your ears and reference professional tracks to make informed decisions when using EQ plugins to remove panned vocals.

Vocal reduction and isolation tools

Audacity's Vocal Reduction and Isolation feature is accessible through the Effect menu and offers several options for removing or isolating vocals. The simplest method involves subtracting one channel from the other, resulting in dual mono output. This technique works on the principle of phase cancellation, where identical audio waves from the left and right channels are inverted and cancel each other out, leaving only the instrumental track.

For a more precise vocal removal, the Split Stereo Track Method can be employed. This involves splitting the stereo track into two mono tracks and inverting one channel. The Low Cut and High Cut sliders can be adjusted to define the vocal range for removal, ensuring that bass lines and percussions are not accidentally removed.

Another notable tool is the Intel OpenVINO Music Separation plugin, which can be downloaded and installed to separate vocals from music within Audacity. This AI-powered plugin offers different separation modes, such as the 2-stem option for isolating vocals and instruments, and the 4-stem version for extracting specific elements like drums and bass.

While these vocal reduction and isolation tools are valuable, it's important to note that the quality of vocal removal can vary depending on the song's mix and the prominence of the vocals. Additionally, as vocals are not always center-panned, these methods may not always yield perfect results, and some artefacts or backing vocals may remain.

Other software programs that offer vocal removal capabilities include AudioKies, Adobe Audition, iZotope RX, and Pro Tools by Avid. There are also machine learning-based tools like Spleeter and Moises, which provide accurate sound separation results.

Phase cancellation technology

Phase cancellation is a technique used to remove vocals or instruments from an audio track by exploiting the concept of phase, which is time-based and determines the cycle of a waveform. Phase cancellation occurs when two waveforms are 180 degrees out of phase with each other, resulting in their signals cancelling each other out.

In the context of vocal removal, phase cancellation is achieved by inverting the polarity of one of the stereo channels, typically the right channel. This is because vocals are usually recorded in the centre of the stereo image, with the same signal present at the same level in both channels. By inverting the polarity of one channel and mixing the left and right channels together in mono, the identical signals in both channels cancel each other out, causing the vocals to disappear.

It is important to note that this technique may not completely remove the vocals, especially if there is vocal reverb or panned vocals present. Additionally, other centrally panned instruments, such as bass and kick drum, will also be removed in the process. To retain these elements, a high-pass filter can be applied to one channel before mono summing, allowing the bass parts to be mostly preserved.

Phase cancellation can be performed using various audio editing software, such as Audacity, Adobe Audition, and AudioKies. These tools provide features like Vocal Reduction and Isolation, which allow for the removal or reduction of vocals from a track.

Frequently asked questions