Audio restoration software are tools which, in theory, we should never need. After all, we all record audio at the correct level in ideal locations without any technical defects or outside influences… Meanwhile in the real world, we are presented with a wide range of audio that is recorded too high and distorted or recorded too low so that when we boost, it is noisy. We end up with location recording with noises that we don't want or spill from adjacent sources. Maybe it was too windy, or the mic rubbed on the person's clothing. Or maybe the microphone was too far away from the source. These are just some problems we turn to audio restoration software to help us resolve.

We last looked at this in 2021. In the last 2 years, the audio restoration market has changed significantly, especially with the addition of artificial intelligence with a number of small new developers producing products that challenge the well-known brands. In this updated article, we highlight a selection of audio restoration software, some stand-alone, some plug-ins and some that offer both options that we think you should try out.

Let’s start with the two brands that historically have been at the top of this sector, iZotope and CEDAR…

iZotope RX

Trusted by the top audio pros around the world, iZotope RX has been built on years of extensive research in advanced digital signal processing and machine learning.

Some of you know me as ‘Mr RX It’, from when I produced a series of videos showing how to use iZotope RX 2 to treat a range of problem audio examples. Fast forward, and iZotope has added some amazing features to RX, especially with the machine learning based tools that opened up a range of possibilities that weren’t available before. 

RX 6 brought cutting-edge technologies for post-production, like lav mic de-rustle, dialogue isolation, and intermittent low-end wind rumble removal. It also introduces creative approaches to age-old music recording problems like microphone bleed, sibilance, mouth clicks, and breaths. 

With RX 7, iZotope added Dialogue Contour to change the intonation, a dedicated dialogue De-reverb module, multi-channel support up to 7.1.2, Repair Assistant, which uses AI to analyse the audio and offers options for you to consider and Music Rebalance, which uses source separation, identifying vocals, bass, percussion, and other instruments, enabling you to be able you to change the mix after the event.

iZotope RX 8 added Spectral Recovery, which restores lost frequencies above 4 kHz, turning bandwidth-limited audio into clearer and more articulate recordings, and a dedicated Guitar De-noise module, which, although refined for removing hums and buzzes from guitar tracks, is also great for any hum and buzzes you might have on any type of audio. RX 8 Advanced brought a Wow and Flutter module to correct pitch variations and fluctuations associated with tape, vinyl and optical transfers.

All of this has given me a huge respect for the team at iZotope and enabled me to make a living helping people improve their audio. But it has to be said that all is not rosy in the iZotope garden. Both RX 9 and RX 10 were disappointing in their new features, with iZotope taking their eye off the ball together with a range of smaller developers now taking up the audio restoration mantle and leaving iZotope behind in both product development and the quality of the results.

iZotope RX 10 Pricing

• RX Elements - $99

• RX 10 Standard - $399

• RX 10 Advanced - $1199

CEDAR

We are going to focus on CEDAR’s software products aimed at our sector, which are CEDAR Studio and CEDAR VoicEX.

The Studio range is made up of eleven processes, all designed to enable you to eliminate a wide range of common problems and significantly improve the sound quality of your audio. Each CEDAR Studio process has been derived from their multi-award-winning CEDAR Cambridge flagship product and aims to provide unsurpassed results quickly and with the minimum of fuss and user intervention.

CEDAR Studio DNS includes both DNS One and DNS Two for all of your dialogue noise suppression requirements in post-production and elsewhere.

If you work with film dialogue, the speed, flexibility, and ease of use of DNS One provides solutions to audio problems that you could not previously solve. It's a standard for multi-channel post-production in the film, video, and TV industries. It's designed to remove motor noise from recordings, eliminating electrical interference and helping to clean up recordings suffering from unfavourable acoustics and poor microphone locations.

In almost all cases, using DNS Two is as simple as switching on LEARN and then adjusting the attenuation to obtain the desired amount of noise reduction.

CEDAR Studio Restore comprises no fewer than eight advanced audio restoration algorithms…

• Declick offers improved impulsive noise detection and a better interpolator, designed to offer superior performance across a wider range of materials than ever before. 

• Decrackle boasts an ability to dig into the damaged signal to identify and remove all manner of ground-in and grungy crackles without damaging the wanted audio. It also removes many forms of buzz and some amplitude distortions from material ranging from cylinder recordings to current broadcasts contaminated with lighting buzz.

• Manual Declick is for restoring the extended scratches and clicks encountered on many audio media - film, tape or disk, whether analogue or digital. Its dual algorithms, optimised for long clicks and scratches, will remove noises such as those found on optical soundtracks, as well as drop-outs and extended digital errors.

• Dethump has been created to eliminate the extended low-frequency thumps that cannot be restored using conventional declicking processes and filters. It replaces the unwanted sound with restored low-frequency audio and is the ideal tool for removing many of the previously intractable problems associated with optical soundtracks, cylinders, 78rpm discs, and even modern recordings when, for example, microphones and stands are bumped.

• Auto Dehiss has an advanced algorithm that has been designed to be far less susceptible to unwanted side-effects, sometimes called twittering, glugging, the 'underwater sound', and even 'space monkeys'. A manual mode is also offered for fine-tuning the results when wanted.

• Debuzz has been designed to quickly remove all manner of buzzes and hums with fundamentals as high as 500Hz. It will track wandering tones and, unlike traditional filters, can successfully restore the audio without unwanted side effects such as limited bandwidth or the hollow sound introduced by comb filters.

• Declip 2 removes clipped samples and reconstructs the original signal, improving clarity, eliminating distortion, and restoring the original dynamic range of the damaged audio.

• Adaptive Limiter employs another algorithm developed by CEDAR that constrains the peak level of the output while retaining the integrity of the input and, with its advanced resampling and noise shaping capabilities, is ideal for mixing, final mastering and other creative duties.

CEDAR Studio Retouch, first seen in 2002, is arguably the product that changed the audio industry and could be the most effective way to deal with otherwise intractable audio problems. Nowadays, it allows you to mark and process complex areas in its spectrogram using the types of tools commonly found in powerful photographic and image manipulation software. This makes it possible to identify all manner of sounds and noises, which can then be manipulated using any of the processing tools at your disposal.

Retouch lives within a plugin window inside your DAW of choice. It takes a little adjusting to get used to the interface and workflow, but once you do, not having to roundtrip every clip really does speed up the process on a day of heavy spectral repair work.

Check out our article How Does CEDAR Audio Studio 9 Compare Against iZotope RX? to learn more about their Studio 9 range of plugins.

VoicEX - In September 2023, CEDAR announced its next iteration in software audio restoration tools, the VoicEX.

“CEDAR's VoicEX voice extractor is close to being a magic button, and it's so simple to use that novices and experts alike can achieve results that would have been unthinkable just a few months ago. The results that users can obtain from this are remarkable, and it sets a new standard for real-time noise reduction.”

Rather than try to gauge the noise level at each frequency and suppress this, with VoicEX, they have chosen to train a neural network algorithm to identify the voices in the signal and create two audio streams - one containing just the voices and the other containing everything else. This approach has been designed to make VoicEX very simple to use. With VoicEX, you will find just two knobs - one to determine the loudness of the background and the other to control the loudness of the speech. Consequently, when you reduce the background, unwanted sounds like noise or excessive reverberation are reduced.  If you reduce the speech control, the voices go away, giving you a ‘clean’ background.  

“Increase either to emphasise that part of the signal. It's incredibly simple to use and takes no time at all to master, but the results can be truly amazing.”

CEDAR Studio Pricing

Unfortunately, CEDAR does not publish the prices of their products. Instead, they have this statement at the bottom of each product page…

“To enquire about the price and availability of CEDAR Studio or to arrange a trial licence, please provide your details below:”

You have to use the Contact page, email them or reach out to a CEDAR dealer to get the prices. In 2021, the price of the Studio Complete bundle with all 11 tools together in one package the price was $4,999. In addition, it is possible to buy the Adaptive Limiter online from the CEDAR website, and the price is $440.

We have also had it confirmed that the price of VoicEX is £750 plus local taxes.

Big And Little

There is no doubt in my mind that iZotope has dropped the ball on being at the cutting edge of product development, instead choosing to do as little as possible to continue to bring in revenue, which is a classic strategy when brands are taken over by venture capital as we have explored in a number of articles.

When it comes to CEDAR, there is no doubt that they still top the forensic market as well as the live, real-time noise reduction markets, both of which can still command the high prices that they charge. Interestingly, CEDAR has taken a change in direction with the release of VoicEX in design but still appears to be sticking to its high-pricing model.

So, what of the little guys? How are they doing? Have they overtaken the likes of iZotope and CEDAR?  Let's take a look…

Absentia DX

The underlying aim of this plugin is to analyse production dialog recordings and then remove obvious hums, wireless rings, and ticks while maintaining the integrity of the human voice. However, the latest iteration of Absentia DX sees a change to a subscription model and an extension to the features. It’s no longer a one-trick pony. 

Standalone Or Plugin - You Choose

The new V3 plug-in includes all the tools from the Standalone app and a local library that makes processing 3x faster. The Hotkeys shortcuts will speed up editing. You can use the Standalone Application to Batch Process the complete sound dailies folder or use the Pro Tools plug-in when editing or mixing to make deeper processing to some clips.

Absentia DX contains 8 different tools…

• Broadband Reducer can now process deeper than before with a lower noise peak model.

• Hum Remover automatically removes hum from recordings. The Hum Remover Smooth setting is designed to be safe for all dialog.

• Air Tone Generator enables you to replace dialogue with room tone while maintaining production moves.

• Doppler Remover enables you to remove hums or whistles that vary in pitch.

• Cicadas Reducer reduces cicadas in the background, leaving the dialog untouched.

• Birds Reducer reduces bird chirps in the background, leaving the dialog untouched.

• Phase Synchronizer enables you to cut Boom and Lavalier together and avoid phase cancellation issues.

• Sonogram Player lets you view the Spectrum of a poly-wav file and play it channel by channel. 

There are two other features which aren’t in the audio restoration workflow but are a great help to dialog editors…

• Transcribe converts dialogue recordings into text, whether it's production sound, ADR, loop group sessions, voice-overs, or any voice recording. Search for alternative lines or specific words and expressions in a user-friendly spreadsheet or directly within your sound database app.

• ALTS Finder enables users to navigate through your transcribed dialogue, listen and compare different takes, and spot the selected takes directly into Pro Tools.

Absentia DX Pricing

There are two options, both subscription models…

• Server is for users who don't need to carry the license with them regularly. The license is activated on your machine, but you can request to move it. Absentia DX Server costs $120 for the first year and then $40 for subsequent years as long as you renew it before the license expires.

• iLok is for users who work on different facilities and need to carry the license around, so the license is activated on a 2nd Generation Pace iLok2 or later. Absentia DX iLok costs $150 for the first year and then $50 for subsequent years as long as you renew it before the license expires.

Accentize

Accentize has 5 products in the audio restoration sector - VoiceGate, DeRoom Pro, SpectralBalance, DialogEnhance and dxRivive. 

VoiceGate is a machine-learning-based audio plugin which can be used to reduce noise from speech and vocal recordings in real time. An artificial neural network has been trained on more than 100 hours of audio data to learn the characteristics of human speech. The separate controls for steady and impulse noises provide parameters to tackle different kinds of noises without introducing many artefacts. In 2022, Accentize released VoiceGate 2, which added an automatic noise reduction option, so you don’t need a noise sample to teach it with. In addition, the upgrade brings a difference listening mode so you can identify what the plugin is removing. Accentize said…

“It has been two years since the release of our first plugin VoiceGate. We have used the past few months to completely rework its audio engine and trained a new neural network which achieves a far more accurate noise reduction for any kind of dialogue recording.”

DeRoom Pro is a machine-learning-based reverb removal audio plug-in built for professional use. An artificial neural network has been trained using many different room scenarios in order to be able to separate direct sound from reflection components. The real-time processing scheme provides instantaneous feedback while changing parameters and allows seamless integration into your workflow without having to switch applications.

In 2022, Accentize released the second generation DeRoom Pro, featuring an improved filtering algorithm which provides higher accuracy when detecting room reflections. This is especially noticeable when used with recordings in larger rooms with long reverb tails. Mike tested the first iterations of VoiceGate and DeRoom Pro back in 2020. Mike said…

“Both DeRoom Pro and VoiceGate will stay on my machine and will be part of my audio restoration toolkit alongside iZotope RX. In fact, it will be installed on my laptop, too, because Accentize allows each purchase to be installed and licensed on 2 computers at the same time.”

SpectralBalance is an intelligent equalizer plugin for dialogue recordings. By continuously analysing the audio signal, it automatically adapts its EQ curve to correct spectral imbalances in order to achieve neutral-sounding speech. The workflow makes it a useful and time-saving tool for professionals who have to deal with many different dialogue takes with individual spectral characteristics. Additionally, SpectralBalance lets you define custom EQ targets, which makes it a valuable helper for ADR matching.

• SpectralBalance can reduce the time spent repetitively listening to dialogue takes and finding the appropriate EQ settings.

• Use dynamic mode to apply individually adjusted EQ corrections. The internal algorithms listen and, if necessary, adapt the processing over 50 times per second.

• Let SpectralBalance listen to a few seconds of target audio and automatically create an individual target EQ curve. This way, you can define how the output should sound for EQ matching tasks like, for example, ADR.

• The core technology of SpectralBalance is a combination of different artificial neural networks, which are all optimized to keep the required processing power to a minimum so you can run multiple instances in parallel on an average workstation.

DialogueEnhance is an intelligent automatic speech-processing tool. Aimed primarily at video editors and podcast creators, DialogueEnhance is a plug-in which offers four modules: noise reduction, dynamics control, spectral correction and loudness boost. Each of these is designed to be easy to use and to allow you to achieve good results with minimal effort. They can be activated or bypassed individually as needed. Much of the processing is achieved automatically with the help of Machine Learning algorithms. 

• The noise reduction module is designed to detect and reduce unwanted background noise in dialogue recordings.

• The dynamics control module is an automatic compressor. In automatic mode, it will determine the compression settings based on the audio input signal. In static mode, once DialogueEnhance 2 establishes a convenient setting, then you have some control over the compression. 

• The spectral correction module will automatically detect and correct spectral imbalances in the audio. This is, in essence, an automatic 3 band EQ consisting of high and low shelving filters and a peak/dip curve. Once again, you can switch to static mode once a satisfactory setting has been achieved.

• The final module, loudness boost, allows you to set a target loudness level in LUFS. Dialogue Enhance 2 then adjusts the output level to achieve the specified loudness. As with the Dynamics and Spectral Correction modules, Loudness Boost has a static mode which allows you to set a fixed gain boost setting if required.

If you want to get a feel for how DialogueEnhnace works, then check out Paul Maunder’s article and video Accentize DialogueEnhance 2 In Action.

dxRevive is a new speech restoration plugin from Accentize, released in September 2023. Utilising groundbreaking AI technology, dxRevive aims to transcend the boundaries of conventional noise removal, working to restore clarity and tone to audio recordings in ways previously unimagined.

dxRevive presents itself as a versatile plugin dedicated to enriching the quality of various dialogue recordings. Unlike many other restoration plugins, dxRevive doesn't merely filter the signal. It goes beyond identifying and reintegrating missing frequency components, yielding studio-like recordings from diverse source materials. Its capabilities span:

• Noise removal

• Reverb elimination

• Restoration of absent frequencies

• Elimination of Codec Artifacts (e.g., from Skype or Zoom recordings)

• Recovery of clipped audio

• Seamless replacement of missing samples to address audio dropouts

• Application of spectral corrections

dxRevive is available in two versions: Standard and Pro.

• The standard version of dxRevive provides a very simple interface which displays a scrolling audio waveform. The original audio is indicated with an outline, and the processed audio is represented with a filled-in waveform.

• dxRevive Pro introduces additional features, namely the option to split the audio into as many as 4 frequency bands, each of which has its own threshold control so you can adjust the amount of processing per band. This is particularly useful in situations where the noise is most prominent in a certain frequency range, and you want to process it more heavily than other parts of the spectrum.

Paul Maunder tested dxRevive and dxRevive Pro. Check it out in his article and video Accentize dxRevive Tested - Perhaps The Only Dialogue Cleaner You'll Need.

Accentize Pricing

• VoiceGate $149

• DeRoom Pro $259

• SpectralBalance $149

• DialogEnhance $69

• dxRevive $99 (Intro offer until Sept 30th 2023 - $74)

• dxRevive Pro $299 (Intro offer until Sept 30th 2023 - $225)

• Complete $1196 (Intro offer until Sept 30th 2023 - $838)

Acon Digital

Acon Digital have 5 products in the audio restoration sector: the Restoration Suite, Extract:Dialogue, DeVerberate, DeFilter and Acoustica, which is an all-in-one mastering and audio restoration bundle offering a similar experience to iZotope RX.

Restoration Suite 2 has four plug-ins for audio restoration and noise reduction - DeNoise, DeHum, DeClick and DeClip. The plug-ins are available as VST, VST3 or AAX for both Windows and Mac or as AU plug-ins for Mac. There are 32 and 64-bit versions for Windows. The Mac version is 64-bit and runs natively on both ARM (such as Apple M1) and Intel processors.

With the second generation release, Acon Digital has improved 3 of the 4 plugins. With DeNoise 2, they have added dynamic noise profiles that help reduce noise that varies randomly over time, such as wind noise or rustle from lavalier microphones, effectively adding DeWind and DeRustle to the Denoise plugin.

With DeHum 2, they added support for Mid/Side (M/S) processing, which can reduce unwanted fluctuations in the stereo image. In addition, the hum detection algorithm now has a new Scan button that automatically fine-tunes the fundamental hum frequency.

Like DeHum 2, with DeClick 2, they have added Mid/Side (M/S) processing, designed to reduce unwanted fluctuations in the stereo image.

Extract:Dialogue separates dialogue from common types of background noise such as wind, rustle, traffic, hum, clicks and pops. The algorithm works in real time and is based on machine learning. Acon Digital has trained it on thousands of high-quality voice recordings and an extensive set of common noise sources to enable artificial intelligence to distinguish dialogue from noise without user interaction automatically. All this AI aims to make Extract:Dialogue extremely easy to use – just add it to your dialogue bus and let it do its magic. With the release of v1.2 and then v1.5, Acon Digital has significantly improved the results each time. When Paul Maunder tested it earlier this year, he said…

“In use, I found Extract:Dialogue version 1.5 to give remarkable results, notably improving on the previous version and achieving a greater amount of effecting noise reduction whilst leaving the dialogue intact. The improvements Acon Digital have made are evident in the quality and naturalness of the processed voice and the new AI model can correct for phase distortions and destructive interference. Check out the video to see and hear it in action!”

DeVerberate is a reverb reduction plug-in that can attenuate or boost the reverberation already present in recorded material. The new version includes a new algorithm based on deep learning for fully automatic reverb reduction of recorded dialogue. By training a neural network on thousands of high-quality voice recordings and a wide variety of acoustical surroundings, the AI can automatically separate dialogue from reverb.

Because removing reverb from an existing recording is a key part of audio restoration workflows, we undertook a set of blind listening tests. You can read more in our article De-Reverb Listening Test Round 2 - The Results in which we pitted the top two finishers from round 1 – Acon Digital DeVerberate 3 and Accentize DeRoom Pro 2 – against iZotope's RX Dialogue De-Reverb. The Acon Digital plug-in was voted as giving the best result in 3 out of 4 of the examples.

DeFilter is a tool designed to even out tonal imbalances in a recording or reduce resonant peaks or dips in the frequency spectrum. Even severe issues such as comb filter effects or standing waves from untreated rooms can be addressed. DeFilter can match recordings to target profiles such as music or speech or alternatively extract the overall timbre from the source – the large strokes – while evening out narrow peaks and dips. The latter effectively removes resonances and comb filter effects while preserving the input signal's tonal character.

Acoustica Premium Edition is designed to be a solution for audio editing, post-production, podcast creation, mastering and audio restoration. Although you can choose between the fully featured Premium Edition or the budget-friendly Standard Edition, to be really useful as an audio restoration tool, you need the Premium version or the Acoustica Post Production Suite. Either way, you get a fast and intuitive audio editor with an extensive set of high-quality processing tools and support for all common plug-in formats.

The Premium Edition adds advanced features such as spectral editing, powerful retouch algorithms, and multi-channel support up to Dolby Atmos 7.1.2. You also get the Restoration Suite 2 and Extract:Dialogue, and new in version 7.4, they have added a DeBird plugin, a DePlosive plugin, and a dedicated dialogue DeClick plugin turned to handle mouth clicks and lip smacks.

There is also ARA2 support included in the Premium Edition, as well as a Transfer plugin in the AAX format offers a quick and easy way to transfer audio clips for processing from Pro Tools to Acoustica – and back again, which is helpful because there still isn’t ARA2 support in Pro Tools for Acon’s products yet.

Paul Maunder spent some time with the Premium Edition, and you can learn how he got on in our article iZotope RX User Tries Acon Digital Acoustica. Paul said…

“The suite of noise reduction tools is very effective, especially for dialogue, so if you’re working in post-production, particularly on a limited budget, Acoustica should definitely be something which you should consider. Suffice to say, Acoustica represents superb value for money, and it exceeded my expectations.”

Acon Digital Pricing

• Restoration Suite $99.90

• Extract:Dialogue $99.90

• Digital DeVerberate $99.90

• Acoustica Premium Edition $199.90

• Acoustica Post Production Suite $299.90

Hush

Hush is powered by machine learning, with Hush and Hush Pro designed to set a new standard for dialogue repair. With Hush, you are able to suppress room reflections, ambient noise, and transient sounds. Unlike most AI-based audio restoration tools, Hush, like dxRevive, doesn’t upload your precious audio to the cloud; they do their magic locally. Hush takes that a step forward and has been designed to only work on Apple Silicon powered computers so that it can make use of the Neural Engine integrated into Apple’s SoCs. Let’s take a look at the features built into Hush…

• Mute all the noise. Hush filters out broadband sounds like ventilation, appliances, traffic, and wind. Restore audio from imperfect environments and produce studio-quality recordings wherever you are.

• Make every room anechoic. Reduce room reflections and comb filtering — from untreated condos to massive auditoriums. Crank up the dial to absorb every trace of reverb, or mix in some dry signal for a more natural sound.

• Dechirp. Debark. Dehonk. Most noise reduction algorithms struggle to remove transient sounds like chirping birds, barking dogs, or honking horns. Hush eliminates them with surgical precision, saving you hours of manual editing.

• Automate your workflow. Hush accepts individual audio files or whole batches, making short work of dialogue stems, voiceover auditions, or audiobook chapters. Choose an output folder and let Hush clean up all your audio in one go.

• Stay cool with Apple Silicon. On M1 and M2 Macs, Hush uses the energy-efficient Neural Engine, leaving your chassis cool, your fans off, and your CPU free for other tasks. It’s also blazing fast, processing an hour-long podcast in a matter of minutes.

• Own it for life. Neural networks with this much power usually require a server — along with hefty usage or subscription fees. Instead, Hush runs entirely on your Mac. Your data stays private, and the app is yours forever.

We last looked at Hush back in April 2023, when we compared three AI Noise Reduction Apps - Hush v Descript v Adobe; Hush was only available as a standalone app. Since then, Hush is also now available as an AAX AudioSuite plugin called Hush Pro.

This plugin is designed for cleaning up dialogue. Optimized for post-production workflows, Hush Pro includes a refined AI model, a mixer-style UI with separate faders for reverb and noise, and support for real-time previews. The plugin comes in two different flavours, integrating seamlessly with Pro Tools:

Hush Mix lets you rebalance speech, reverb, and noise as if they were separate channels, previewing the results in real-time. You can control the precise signal-to-noise ratio with decibel-based faders, link channels for faster edits, or toggle the mute, solo, and bypass buttons to make a/b comparisons on the fly.

Hush Split uses the same engine under the hood but prints speech, reverb, and noise into separate outputs. You can tweak the clip gain on each lane, add other plugins, or make non-destructive edits (for example, isolating a section of room tone to fill silence elsewhere in the session).

Hush Pricing

• Hush (Standalone) $89

• Hush Pro (AAX AS Plugin) $249

Supertone

Earlier this year, we announced that Supertone had released a beta version of GOYO Voice Separator. GOYO is an AI-powered plugin that offers the user the ability to decrease or increase noise, reverb, and ambience using three simple knobs.

Then, at the end of August 2023, Supertone announced their plans for the official release of Supertone Cleary, which will be their official product, replacing GOYO. The upgraded plug-in has been designed to retain Goyo’s simplicity and powerful algorithm, together with additional features like enhanced CPU performance and offline workflow. The approach differs from some other AI de-noise, de-reverb solutions in that it offers real-time control of the three elements, which means the user can adjust the three elements to taste. Our early tests suggest it offers some impressive results. 

Supertone has stressed in their pre-release communications that Supertone Clear will work completely offline, with the only exception being when you need to activate and deactivate the plug-in.

There will be a free trial available once Supertone Clear has been released.

Supertone Pricing

• Supertone Voice Clarity will be available for purchase starting in early October for $99. 

Waves

Waves have 12 products in the audio restoration sector, but for 8 of them, it could be argued they are legacy products as they haven’t been updated in a long time. However, that doesn’t mean Waves has been quiet on the audio restoration front, far from it. They have developed the Vx Clarity range, and it’s that range we will focus on in this article. Back in 2022, Waves released Clarity Vx and Clarity VX Pro. The two products share the same neural network engine, which Waves has trained to recognise vocals and separate them from noise and are based around the same one-knob operation, making it easy to choose the vocal-to-noise balance.

Clarity Vx is the simpler model and has 3 neural networks.

1. Broad 1 preserves both main and background voices when the recording has more than one voice.

2. Broad 2 is recommended for severe noises and for separating a main voice from secondary background voices.

3. Broad ECO is a low-CPU neural network for dialogue only, great for heavy sessions where you need to use many instances of Clarity Vx.

There is a really useful stereo mode. Clarity Vx knows that stereo’s two channels can differ, so a special Double mode analyzes each channel separately and then processes them appropriately. Of course, Clarity VX can also process mono audio.

If you want to know more about Clarity Vx, then check out our article and video, Waves Clarity Vx Realtime Noise Reduction Tested For Music. In his conclusion, Luke said…

“If you don’t want artefacts in the cleaned up audio, or tonnes of complex functionality to navigate, Clarity Vx makes a very interesting proposition indeed

Clarity Vx Pro is the higher-specced model. It has 4 neural networks.

1. Broad 1 is better at preserving both the main and secondary voices.

2. Broad 1 HF is a variant of Broad 1, trained to better preserve high frequencies.

3. Broad 2 is recommended for severe noises. It is also better at separating the main voice from secondary background voices.

4. Broad ECO is a low-CPU network recommended for heavy sessions where many instances of Clarity Vx Pro are needed.

With Clarity Vx Pro, not only can you remove the noise to leave just the voice, you can go the other way to create ambience fills that you can use under ADR by turning the main control all the way to the left.

Clarity Vx Pro also allows you to split the audio into 6 bands to process different frequency ranges, each with independent processing amount, solo, bypass, delta and pure gain controls. There is also VCA control over all the bands to adjust global processing amounts, maintaining the ratio of the multiband settings.

There is also a Reflections control to restore the natural reflections of the voice after noise reduction.

If you want to learn more about Clarity Vx Pro, then check out our article Waves Clarity Vx Pro - First Look At AI Noise Reduction For Post.

Damian Kearns said in his conclusion…

“Clarity Vx and Clarity Vx Pro are without a doubt the greatest pair of real time dialogue noise reduction software plugins I have ever used. Being able to monitor and automate through Clarity Vx or Clarity Vx Pro in real time helps me maintain focus on the dialogue editing at hand, as I’m no longer left wondering how things will sound -and what my gain structure and signal processing will be- after I finish editing.

Clarity Vx offers identical results to Clarity Vx Pro, owing to the fact they have the same Waves Neural Networks. I tend to use Clarity Vx Pro when I’m really trying to clean up particularly tricky dialogue using the advanced settings.”

In May 2023, Waves extended its Clarity Vx product line with the release of Clarity Vx DeReverb and Clarity Vx DeReverb Pro plugins. They follow the same format in that they are based on a one-knob design and have dedicated neural networks for speech and singing.

Clarity Vx DeReverb is Clarity Vx DeReverb Pro’s little brother – it has the same innovative Neural Network engine but without the advanced features.

Clarity Vx DeReverb Pro is a Neural Network powered dereverberation plugin, able to reduce and remove reverb from both dialogue (speech) and sung vocals. With the same intuitive operation of the first Clarity, DeReverb Pro works instantly with the big knob whilst also allowing precise and flexible control over the processing across the frequency spectrum.

Six continuous bands give you the ability to smoothly shape and contour the reverb depending on the source material and production. Clarity Vx DeReverb Pro also works in real-time; it is automatable and does not require any file bouncing or outside clients to send/receive files.

If you want to know more about these two dereverb tools, then check out Paul Maunder’s article Waves Clarity Vx DeReverb And DeReverb Pro Tested.

Waves Pricing

• Clarity Vx $149

• Clarity Vx Pro $799

• Clarity Vx DeReverb $99

• Clarity Vx DeReverb Pro $249

Zynaptiq

Zynaptiq has taken a different approach in not producing a wide range of audio restoration tools; they have chosen to target a number of specific problems with tools like UNVEIL to handle excessive reverb, UNFILTER to handle excessive resonances and to match EQ curves and UNCHIRP to handle artifacts caused by lossy audio encoding & other FFT processes, like spectral denoising.

UNVEIL is described by Zynaptiq as a de-reverberation and signal-focusing tool. This makes it invaluable for mixers who need to dry up over-ambience in a recording, be that an individual track or a whole mix. The plugin works in real-time and can also be used to modify the characteristics of the ambience as well as reduce it overall. Zynaptiq adds that: “UNVEIL is designed to leave discrete echoes untouched. Feedback delay is often a relevant part of a composition and should thus be considered as part of the signal rather than as reverb.” Uses cases for UNVEIL include:

• Audio engineers can attenuate reverb in recordings

• Film mixers and dialog editors can remove reverb from dialog and location sound

• Music producers can turn vintage drum loops rich in reverb into dry recordings and vice-versa

• Foley artists can remove ambience from sub-optimally captured sounds

• Mix and mastering engineers can bring key mix elements into focus by attenuating components commonly referred to as "mud"

To learn more about what UNVEIL can do, check out our article and video, How To Fix Reverberant Voices Fast.

UNFILTER from Zynaptiq is the plugin that can deal with dips arising from comb filtering and also can even restore frequencies in other sources with seemingly little information left to work with, making UNFILTER an extremely powerful utility for post-production and music mixers alike and has a unique feature set that makes new ways of working possible:

• Automatically detect and remove resonances, equalization, roll-offs and the effects of comb filtering from musical, location, dialog and surveillance recordings.

• Apply the measured filter response to other signals to place them in the same "acoustic world".

• Export the measured filter response to or import it from a linear- or minimum-phase impulse response file (WAV).

• Mastering grade adaptive, free-form and graphic equalization.

• Extremely steep 96dB/octave shelving high-pass filter.

• Output limiter for unsupervised use in batch processing applications.

• Highly time-efficient workflow and UI

To learn more about how to remove colouration caused by comb filtering using UNFILTER, check out our article and video, Reduce Dialogue Colouration In Minutes.

To learn more about how to match EQ using UNFILTER, check out our article and video, Is This The Quickest Way To Match EQ When Mixing?

UNCHIRP is a plugin that is designed to remove artifacts caused by lossy audio encoding & other FFT processes, like spectral denoising. Zynaptiq says that you can reduce musical noise and high-frequency chirping (a.k.a. “underwater sound” or “warbling”) and restore lost transients & treble.

If you need to salvage poorly encoded music, speech that’s gone through a low bit rate uplink, over-denoised audio, or footage recorded with consumer mobile devices, then you might want to check out UNCHIRP.

To learn more and download a demo, go to the UNCHIRP page on the Zynaptiq website.

Zynaptiq Pricing

• UNVEIL $359

• UNFILTER $359

• UNCHIRP $359

• Repair Bundle $749

Honourable Mentions

There are a number of tools that we feel don’t make the cut for the main part of the article but nevertheless are worthy of a mention…

Descript

Descript is the company behind some smart audio transcription software, or should we say very smart. In a very short space of time, they’ve created a platform that can record video, screen capture, audio, transcribe it and do lots of clever stuff, often with just the click of a button.

What Descript are aiming to do is help podcasters, video tutors, and a whole host of other people who record video and audio make the best possible content. In July 2021, we discovered Descript and couldn’t believe how good it was.

We had been using Descript for some time as a transcription tool at this point, but when they emailed us in 2021 announcing Studio Sound making comparisons to software like iZotope RX, we had to check it out. For our test, we decided to be brutal. We created some challenging tests, including recording in a kitchen on a MacBook Pro using the internal microphone.

At the time, we said, “It would also be nice to have some adjustment as, at times, it sounds overcooked”. As you can see from the screenshot of the current version above, that is exactly what Descript did, by adding an Intensity control.

Descript Pricing

Descript has a subscription pricing structure starting with a free option, but not you can only process files with Studio Sound with a duration of up to 10 minutes with the free version. If you need to process longer files then you will need to move onto a paid plan.

Adobe Enhance Speech

Adobe’s aim is not dissimilar to that of Descript, to help podcasters not have to worry too much about the technology and instead concentrate on what is being said so that people can edit words, not waveforms, by offering Adobe Premiere Pro’s speech-to-text technology in Adobe Podcast to enable users to edit audio using a transcript easily.

Everything is online. There is no app to download. Just go to the dedicated Adobe Podcast page, sign up to get access, and you are away.

As part of Adobe Podcast, Adobe has added Enhance Speech, which has been designed by Adobe to “increase clarity by removing background noise and sharpening your voice’s frequencies. It makes it sound as if everything was recorded in a professional studio”. There is nothing to do other than upload a file using the browser interface, turn on ‘Speech Enhance’ and then download the result.

Enhance Speech Now In Premiere Pro

At IBC 2023, Adobe announced the first professional iteration of Speech Enhance by embedding it into the beta version of Adobe Premiere Pro as a new feature.

Like Descript, Adobe has added a Mix Amount control so you can adjust the amount of processing.

The other good news is that unlike Adobe Podcast, which is hosted in the cloud and requires you to upload your audio to the cloud to be processed, the Premiere Pro feature has been optimised for use on your computer and doesn’t need an internet connection.

Interestingly, although Premiere Pro will work on Intel-powered Macs as long as they have a 7th Gen or newer CPU, Adobe states that “M1 and M2 have significantly better performance than Macs with Intel processors.” and for Windows machines, they say that “The performance will be significantly better with recommended spec graphics cards.” Also, be aware that the Enhance Speech feature in Premiere Pro only supports mono and stereo audio files. It doesn’t support multichannel audio or nested sequences.

Enhance Speech Pricing

When Adobe Podcast was released, it was a free beta, and at the time, we had no information as to how it would be priced. With Enhance Speech now being added to Premiere Pro as a new feature at no extra cost, Adobe’s plans for Enhance Speech are becoming clearer. It may well be that they will add it to Adobe Audition in a future update.

Klevgrand Brusfri

Brusfri actually means ‘noise-free’ in Swedish, which perhaps gives an indication of what it can do. It is an audio noise reducer with a simple and straightforward interface. It is well suited for cleaning up noisy audio recordings while retaining sound quality. Unlike some other noise reducers on the market, Brusfri doesn’t mess with audio phasing to suppress noise (a technique that often comes with audible side effects). Instead, multiple fine-tuned gates are used to handle the unwanted noise.

• Several frequency-trimmed expanders build an intelligent noise suppressor

• No phase-altering, which leaves the desired sound untouched (no squishy artifacts as in many other noise reducers)

• Intuitive and unique user interface

To learn more about Brusfri from Klevgrand check out our free tutorial on how to use Brusfri and see if Brusfri From Klevgrand Is A Challenge To iZotope RX? Check Out This Video

Klevgrand Brusfri Pricing

• Brusfri is priced at $59.99

Sonnox Restore

Sonnox Restore is a suite of three plug-ins designed for audio restoration. Its algorithms and novel features are designed to allow for the removal of pops, clicks, crackles, scratches, hums, buzzes and background noise from virtually any recording without damaging the important 'wanted' audio content. It’s a shame that, to date, Sonnox hasn’t improved these tools for quite a while, which is why they have been relegated to the Honourable Mention section.

With their interfaces, Sonnox Restore's plug-ins are designed to be quick and easy to use. The DeClicker has a dialogue mode and 'exclude box' for retaining wanted audio; the DeBuzzer has 3 frequency ranges and real-time tracking for wandering problem frequencies, and the DeNoiser, with its independent de-hisser, offers powerful real-time broadband noise reduction.

• Oxford DeNoiser - The DeNoiser Plug-In, with its integrated de-hisser, delivers broadband noise reduction in real time. The process is based on a continual analysis of the input signal in order to build a noise profile. Alternatively, you can use Freeze Mode to sample a section of audio containing just target noise. Sonnox acknowledges the delicate balance between removing noise and retaining the character of your original recording, so have added a Warmth control to return some richness and added a DIFF function to audition only the removed noise.

• Oxford DeBuzzer - This has been created for fast and effective removal of buzz, hum, whine and whistles. You use the intuitive display to help isolate the fundamental frequency – it constantly gives feedback on the top contenders – or you can define it manually with the help of a tone generator. If the fundamental wanders, the default Auto Tracking Mode is designed to follow it in real time. The harmonics of your problem's fundamental frequency will also be tracked. All that remains is to engage the default comb filter or the optional parametric EQ and then fine-tune the content that’s removed using DIFF auditioning.

• Oxford DeClicker - Declicker has been designed to remove pops, clicks and crackles using three discrete sections. Each has a threshold and sensitivity control to define the number of events that are detected and removed. The unique Events Graph shows the level and duration of each pop, click and crackle as it’s detected in three separate colour-coded groups.  Sometimes, there will be events detected that represent audio you want to keep rather than remove. So Sonnox has built the Exclude Box to refine the number of repaired events and also Dialogue Mode for independent control over speech in the programme.

To learn more about Sonnox Restore, check out our article Sonnox Restore Now AAX 64 Bit Compatible & With A New Price Too.

Sonnox Restore Compatibility And Pricing

You can buy the plugins separately or buy the bundle and save 25%…

• Oxford DeNoiser - $250.03

• Oxford DeBuzzer - $250.03

• Oxford DeClicker - $250.03

• Sonnox Restore Bundle - $561.27

Blind Tests

To help you evaluate the products covered in this article, as well as the links to tests and reviews we have completed for individual products, links to which you will find above in each section, we have also undertaken a number of comparative blind tests, where we asked you to vote on which example sounded the best.

Low Cost Noise Reduction Tools Shoot Out - Results

In 2022, we ran a shootout comparing low cost noise reduction tools. The shootout featured four pieces of problematic audio, each exhibiting a different problem: hum, crackle, clipping and general broadband noise. The noisy audio was processed with five different noise reduction tools:

• iZotope RX Elements consisting of De-hum, De-click, De-clip and Voice De-noise

• Acon Digital Restoration Suite 2 consisting of DeNoise2, DeHum 2, DeClick2 and DeClip2

• Acon Digital Extract:Dialogue

• Waves Clarity Vx standard edition

• Antares Autotune Soundsoap

To learn about the tests and results, check out our article Low Cost Noise Reduction Tools Shoot Out - Results.

Premium Noise Reduction Shootout - The Results

In 2022, we ran a shootout comparing premium noise reduction tools. The shootout featured five pieces of problematic audio, each exhibiting a different problem: transformer buzz, broadband noise from a fan, fluorescent light buzz, hiss and crackle. The noisy audio was processed with four different noise reduction tools:

• Acon Digital Acoustica 7 Premium Edition

• Sonnox Restore bundle

• Waves Clarity Vx Pro

• iZotope RX9 Advanced

To learn about the tests and results, check out our article Premium Noise Reduction Shootout - The Results.

De-Reverb Listening Test - The Results

In 2023, following our listening test comparing some of the most popular de-reverb plug-ins, Paul Maunder took a look at the results and introduced a second shootout for you to listen to and vote on.

The de-reverb shootout featured four dialogue recordings made in different acoustic settings, each of which exhibited an obvious reverb. The recordings were each processed with five different de-reverb plug-ins… 

• Accentize DeRoom Pro 2

• iZotope RX De-reverb

• Zynaptiq Unveil

• Acon Digital DeVerberate 3

• Plug-in Alliance SPL De-Verb

In each case, we were careful to listen to what the plug-in was doing and adjusted the settings to give the best result, factoring in both reverb reduction and audible artifacts.

To learn about the tests and results, check out our article De-Reverb Listening Test - The Results.

De-Reverb Listening Test Round 2 - The Results

In round 2 of our dialogue de-reverb listening tests, which we devised in response to comments from members of the Production Expert community, we pitted the top two finishers from round 1 – Acon Digital DeVerberate 3 and Accentize DeRoom Pro 2 – against iZotope's RX Dialogue De-Reverb.

To learn about the tests and results, check out our article De-Reverb Listening Test Round 2 - The Results.

Dialogue Cleanup - AI Versus Audio Professional - The Results

In March 2023, we ran a listening test comparing the effectiveness of AI/ machine learning noise reduction tools with the results achieved by an audio professional using manually operated software. You can find the original test here.

Four dialogue recordings were included in the test, and each one was processed a total of four times using the following software:

• Adobe Speech Enhance (AI/ machine learning)

• Descript (AI/ machine learning)

• iZotope RX 10 Advanced (Manually operated)

• Acon Digital Acoustica Advanced (Manually operated)

To learn about the tests and results, check out our article Dialogue Cleanup - AI Versus Audio Professional - The Results.

Battle Of The AI Noise Reduction Apps - Hush v Descript v Adobe

In April 2023, we put the three popular AI solutions against one another in a blind test.

• Descript Studio Sound

• Adobe Speech Enhance

• Hush

To learn about the tests and results, check out our article Battle Of The AI Noise Reduction Apps - Hush v Descript v Adobe.

Conclusion

If you have got this far, I suspect you will be wanting to know which one is best. Hopefully, the blind tests we have done and will continue to do will help, but I continue to believe that anyone in the audio restoration business needs to have a selection of different tools. Different in what they do and from different brands. I say this from experience because I have found time and time again that different products handle different audio problems differently. Sometimes, one is better than another for one problem, and then when presented with the next problem that might appear to be similar, say like hiss, a different product will be better. My advice is to download the demos and try them out for yourself on audio you need to clean up and decide for yourself which ones are the ones that are best for you. The good news is that audio restoration is not as expensive as it used to be. Just look at the new kids on the block.

Studio Sound and Adobe Podcast can achieve amazing results but are not really suited to professional workflows and are dependent on uploading content to the cloud, with the associated security concerns. However, with the integration of Enhance Speech by Adobe into Premiere Pro, they have removed the need to upload to the cloud and made the AI work on the host computer.

It is interesting that a number of brands have just announced products that, on the face of it, look very similar, in Accentize with dxRevive, Supertone (formerly known as GOYO) with Voice Clarity, CEDAR Audio with VoicEX and Hush Audio with Hush Pro, together with Waves Vx Clarity range released in 2022 and 2023. They all have a common thread in that they are tools designed to clean up dialog, and rather than have to use a range of tools to achieve this; they are all designed to handle a wide range of problems within one interface.

But for me, one of them stands out, and that is dxRevive from Accentize. That is because dxRevive and dxRevive Pro are not only audio restoration tools but also they are audio reconstruction tools, as are Studio Sound from Descript and Adobe Podcast. Looking at the comments across social media, this reconstruction element seems to be working for them and against them, with people saying of dxRevive that they find the result too good. This, for me, is why I prefer dxRevive Pro because it has two algorithms: Studio, which includes the reconstruction element and Retain, which doesn’t. You have a choice. I also do like the band splitting feature that comes in dxRevive Pro, as it means that I don’t need to over-process areas in the frequency spectrum.

The other takeaway for me is that with the exception of Waves with the Clarity Vx range and CEDAR with the VoicEX plugin, established brands like iZotope and Sonnox are not innovating in the audio restoration sector, and that is a real shame.

Instead, smaller developers like Accentize, Absentia, Acon Digital, Hush Audio and Supertone are the brands at the cutting edge of developing new tools in the audio restoration world, and most are very affordable.