Quality Focused
AI Powered
Vocal Remover
9.0
Extract accompaniment, vocal, backing-vocal, and instruments from any audio
By using our quality-focused AI
AI Powered
ARTIFICIAL INTELLIGENCE
DangoAI is an advanced neural network, trained on a large dataset of songs, that understands various styles of music and can separate anything from any audio
Quality Focused
QUALITY FOCUSED
Tired other software tools that produce muddy, noisy and artifact-filled results? Try our #1 AI algorithms that produce clear and crisp results
Simple and Fast
SIMPLE AND FAST
Zero threshold, one-click upload, rely on our multi-server cluster computing power, no queue required, 1 minute to complete. Fast, simple, and good
8
Introduce
.0 Series
DangoAI Algorithm
Now Available
DangoNet 3
Neural Network
9.0 Series with New Gen-3 DangoNet Benchmark Algorithm.
Unique and independently developed algorithm, superior to similar software worldwide.
Improvements in network architecture lead to performance far exceeding that of previous generations.
Training data increased by 50%, AI more "versatile" and able to understand different styles of music
New WMIR recovery module creates an authentic accompaniment experience.
Quality Focused
DangoAI is quality focused,
and has been recognized by a large number of users in the industry.
Our algorithm is not muddy, no artifacts, no noise. The sound is crisp and clear,
Try it yourself and compare with other software.
+ 0.43 dB
higher SDR (Source Distortion Ratio) than the previous generation.
134 Million
With hundreds of millions of neural network parameters, DangoAI is born for quality.
WMIR 3.0
New WMIR architecture allows the output file to fundamentally improve the output muffled, noisy, and artifact problems.
Better
Outputs
DangoAI cares the authentic listening experience.
Our algorithm was crafted through numerous experiments and auditory feedback, rather than solely relying on tedious academic indicators.
Other Software
12~16 KHz
512 FR
DangoNet Algorithm
22 KHz
4096 FR
*
KHz: Kilohertz. Higher frequency means clearer high-pitched sound and more lossless output.
FR: Frequency Resolution. More audio resolution consumes more computing power, but can focus on the details of the song more finely.
9.0 Series
Controllable
Better output, less noisy
The 9.0 series adopts a new "modular" controllable architecture,
completely solve the problem of AI "learning and forgetting"
AI has stronger fitting ability and is more suitable for various styles of music
New Controllable architecture
Split the original complex NN architecture into sub-modules and process your music together.
73% reduction
In stubborn noise/bleeding residue than previous generation.
0.43 SDR
New architecture brings additional 0.43dB SDR improvement, output is closer to the original sound quality
Learn More
Cares
Sound Quality
9.0 Series brings new triple sound quality protection technology
Compared to other software, DangoAI understands your extreme requirements for sound quality
Dango-Only Audio Fixer
Not only remove, it can also "fix"
Dango-Only Transient Protector
Understand and protect energy changes
Dango-Only Pure Outputter
World #1 zero-artifact output module
Repair damaged accompaniment
Recreate & fill broken tone
Other software only focus on "delete" vocals, but they don't care about how the accompaniment sounds muddy after the vocals are removed.
DangoAI emphasizes the output quality with an AI-based repair module, repaired areas where instruments were damaged by removing vocals.
Recreate the damaged tone and fill it back, restore the crispness of the sound.
Listen & Compare
Other Algorithms
Broken string instrument
DangoNet
Crisp & Clear
Headphones recommended. Click to
Understands
Various styles of music
To extract music of a certain style, AI needs to learn from a large number of similar-style music samples to understand how to extract music of that style
In the 9.0 algorithm, we increased the training data by 50% again, making the extraction process more intelligent
Understand
Chinese Instruments
Dango has uniquely optimized Chinese songs to make them more suitable for the local environment.
Fit Flute, Xiao, Pipa, Erhu, Suona and other Chinese instruments.
Listen and compare
Close
Other alg lack knowledge of the the Erhu and wrongly treat it as human voice, leading to its deletion from the accompaniment. DangoNet can correctly identify the instrument's timbre and preserve it.
Understand
Orchestra
Dango now has a deeper understanding of the composition of various orchestral instruments. And ensures that parts of these are not mistakenly deleted.
Listen and compare
Close
During the interlude stage of music, when only the accompaniment is played, other algorithms may mistakenly identify instruments as vocals, leading to a decline in the overall quality of the accompaniment. Dango AI attention mechanism effectively avoids this issue by correctly distinguishing between human voices and instruments.
Understand
What you should covered
Dango new "Covered Remover" has a better understanding of the composition patterns of complex vocals music. It only eliminates the vocals that need to be covered, but does not eliminate the vocals that do not need to be covered.
Listen and compare
Close
The music producer deliberately "CHOP" the vocals and treats them as instrument, but they do not require elimination or covering. Dango will only remove elements that need to be covered (such as lead or backing vocals), while intelligently preserving elements that do not require coverage (such as chop or choir).
Understand
Complex Vocals
The distinctiveness of operatic soprano, the rich reverberation of human voices in modern music, and sonic manipulations like delay and vocoder all contribute to the immense challenge of removing human vocals from music. Dango enhances its training with an array of richer human vocals, enabling the AI to learn and become more "informed" and easier to remove them from your songs.
Listen and compare
Close
A frequent issue encountered in other algorithms is the disregard of audio phase information, leading to "phase artifacts" during output, manifesting as persisting high-frequency "sizzling" vocal sibilance. Dango cares phase and delivering clean accompaniment without any residual artifacts.
Multiple Algorithms
Not only Vocal Remover
For intricate music scenes, Dango exclusive multi-extraction algorithms
Accompaniment
Extractor
Extractor #1 quality accompaniment
From any music
Backing Vocal
Keeper
Only remove lead vocals
Keep backing vocals
Covered
Remover
Only remove the vocals
which need to be covered
Voice
Cleaner
Extractor made for vocals
More cares vocal quality
Stem
Separator
Extract or remove any instruments
From any song
* Change the algorithm when uploading to use the above algorithms. Default is "Accompaniment Extractor".
Better
And Faster
Dango 9.0 series has more parameters and calculations, but the speed is increased.
We optimized to an extreme level for separation task.
Achieving speeds as fast as 8 seconds for completion.
* During peak periods, high-speed channels may occasionally be unable to allocate resources for you
Feedback Channel
Bad result?
Use the feedback tool and select music clips with poor parts.
Dango will use them to improve the algorithm.
FEEDBACK
Community Incentive (Coming Soon for Non-China regions)
Submitting the dry vocal file of your cover with compensation,
Which will be used to improve Dango separate ability.
ENCOURAGE
Together
Evolve AI
Dango open "Evolution" channel
Regularly update the algorithms