PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchSome links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Yes—you can generate a complete song with lead vocals and accompaniment in WanGP by selecting YuE2‑3B, entering a style prompt, and supplying sectioned lyrics. The practical starting point documented by the YuE2 project is Linux, Python 3.12, and a BF16-capable NVIDIA GPU with 24GB of VRAM. WanGP provides installation options for Windows, Linux, and macOS, but your actual memory requirement depends on its quantization, attention-kernel, and GPU-specific settings.
For the most controllable first attempt, use cot="full". It creates a symbolic melody-and-chord plan and can save an editable score before rendering the audio. Save the entire artifact directory—not just the playable file—so you can inspect the plan, regenerate after edits, and compare settings reliably.
As an Amazon Associate I earn from qualifying purchases.
What does YuE2‑3B do inside WanGP?
YuE2‑3B is an open music-generation model that turns lyrics and a style description into a complete recording with vocals and accompaniment. It first represents the requested song as a symbolic composition plan—melody, chords, and related musical decisions—then synthesizes audio from that plan and its learned acoustic representations.
WanGP adds a local interface and workflow around the model. You choose YuE2 in its audio workspace, let the application download the required checkpoints, enter your prompts, and inspect the resulting audio and saved artifacts. WanGP is the current product name associated with the repository that was previously called Wan2GP.
#1 Best Overall
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
The model can be used in three planning modes. Full planning is generally the best starting point for an original song; melody planning is intended for cover-style work where you want to preserve a supplied melodic contour; and planning can be disabled when you prefer direct generation without a symbolic score.
| Planning mode | Setting | Best use | What you retain |
|---|---|---|---|
| Full | cot="full" |
New songs and maximum composition control | Melody-and-chord plan plus rendered audio; an editable score can be saved |
| Melody | cot="melody" |
Covers or projects where the melody should follow a supplied score | Melodic plan with freer accompaniment |
| Off | cot="off" |
Direct generation when a symbolic plan is not wanted | Rendered audio without the planning stage |
You can also provide ABC notation in supported planning modes when you already have a score and want to control the composition more precisely.
What GPU and software do YuE2‑3B and WanGP require?
The documented YuE2 starting point is a BF16-capable NVIDIA GPU with 24GB of VRAM, running on Linux with Python 3.12. Treat 24GB as the supported baseline described by the project, not as a guarantee that every WanGP configuration will consume exactly that amount.
Hardware baseline
- GPU: NVIDIA graphics card capable of BF16 with 24GB VRAM for the upstream configuration.
- System: Linux is the environment specified by the YuE2 generation guide.
- Python: Python 3.12 in the documented upstream setup.
- Storage: Keep substantial free space for model checkpoints, optional assets, generated audio, scores, token files, and latent data. An NVMe SSD can reduce download and file-access delays, but it is not a substitute for VRAM.
- Network: The first launch may need internet access because WanGP downloads checkpoints and optional assets when a workflow requires them.
WanGP also publishes installation paths for Windows, Linux, and macOS, including a Windows desktop launcher and Pinokio route. Follow the current WanGP instructions for the Python, PyTorch, CUDA, Triton, and attention-kernel combination that matches your GPU generation. Those choices can alter practical memory use and whether a workflow starts successfully.
What the 24GB figure does—and does not—mean
It means the upstream project documents a 24GB BF16-capable NVIDIA card as its supported starting point. It does not promise that every optimization, quantized checkpoint, or WanGP kernel setting needs exactly 24GB, nor does it guarantee that a lower-memory card will run the same workflow. If you are below the documented baseline, check WanGP’s current quantization and memory-saving options rather than assuming a failure is caused by your prompt.
How do I install and open YuE2 in WanGP?
Use WanGP’s current installer or launcher for your operating system, then select the YuE2 audio workflow. The model files are not necessarily present in a fresh installation, so allow the first-run download to finish before diagnosing generation problems.
Recommended Free Tools
- Install WanGP. Use the official script for your operating system, the documented Windows launcher, or the Pinokio installation path. Match the GPU-specific runtime guidance to your NVIDIA driver and CUDA generation.
- Start the application. Give it permission to access the directories where checkpoints and outputs will be stored.
- Open the audio workspace. Select YuE2 from the available audio models or workflows.
- Download required files. Let WanGP fetch the YuE2 checkpoints and any optional assets requested by the workflow. Keep the application open until downloads and integrity checks complete.
- Prepare your inputs. Enter a concise style prompt and lyrics divided into labeled sections such as
[Verse],[Chorus], and[Bridge]. - Choose a planning mode. Use full planning for a new composition, melody planning for a cover workflow, or off when you intentionally want direct generation.
- Generate and inspect. Listen to the result, review the score and metadata if they were saved, and check whether the output was truncated before deciding that the prompt failed.
If WanGP does not list YuE2 after installation, confirm that the checkpoint files are in the expected YuE2 layout, then restart the application. A missing model entry is usually a discovery or file-layout issue rather than a lyric problem.
How should I write the style prompt for a full AI song?
Put musical identity in the style field and keep the first request coherent. Include the genre, instruments, vocal character, language, and tempo; then add production details only when they clarify the sound you want.
Rank #2
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
Use a compact specification
A reliable pattern is:
language, genre, tempo, vocal character, key instruments, rhythm or texture, chorus or mix character
For example:
English, cinematic synth-pop, 112 BPM, intimate female lead vocal, pulsing bass, analog pads, crisp electronic drums, wide chorus, polished modern mix.
This tells the model who is singing, what musical vocabulary to use, how fast the track should feel, and which arrangement elements should be audible. If the first generation is unstable, remove adjectives before adding more: a shorter, internally consistent prompt is easier to diagnose.
Details worth specifying
- Language: State the intended lyric language explicitly, especially when names or loanwords could be pronounced in several ways.
- Genre and era: “Cinematic synth-pop” gives a clearer target than “good pop.”
- Tempo: Give a BPM when the groove matters. It is a target, not a promise that every bar will measure perfectly.
- Vocal character: Describe lead type, intimacy, intensity, or delivery rather than naming a living singer.
- Instrumentation: Name the core palette—such as pulsing bass, analog pads, piano, strings, or electronic drums.
- Arrangement and mix: Terms such as “wide chorus” or “polished modern mix” can guide density and spatial impression, but they cannot replace a clear genre and vocal request.
Avoid putting plot instructions, long prose explanations, and the lyrics themselves into the style field. Keep words in the lyrics field so the model can distinguish musical identity from text to be sung.
How do I format lyrics so YuE2 sings a complete song?
Use explicit section tags and a readable sequence. Tags give the model a structural map, while short, consistently formatted lines make alignment easier.
[Verse]
Neon rain along the avenue
Every window turns the night to blue
[Pre-Chorus]
I can feel the silence start to break
One more breath is all the heart can take
[Chorus]
Stay with me beneath the city light
Turn the falling dark into the night
[Verse]
Signals bloom above the crowded train
Old refrains are running through the rain
[Bridge]
If tomorrow changes where we go
Keep the rhythm only we can know
[Chorus]
Stay with me beneath the city light
Turn the falling dark into the nightSection tags and practical lyric checks
- Use tags such as
[Verse],[Pre-Chorus],[Chorus], and[Bridge]consistently. - Repeat a chorus intentionally when you want a recognizable hook; do not rely on the model to infer a refrain from untagged paragraphs.
- Keep line lengths reasonably similar within a section. Large swings in syllable count can make phrasing feel rushed or leave unused musical space.
- Proofread names, punctuation, and language before generating. A vocal model cannot reliably correct a typo that changes pronunciation.
- Start with one clear song structure. Add unusual tags or spoken passages only after the basic workflow produces a stable result.
Lyrics are not a guarantee of exact syllable-to-note placement. If a phrase is consistently rushed, shorten it, split it across lines, or revise the surrounding section rather than changing several settings at once.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Which YuE2 planning mode should I choose?
Choose cot="full" for an original song when you want melody and chords planned together. Choose cot="melody" when a supplied melody should guide a cover, and choose cot="off" only when you deliberately prefer direct generation without a symbolic composition stage.
Full planning: the default for original songs
Full planning creates a melody-and-chord plan and can save an editable score. That score gives you a control surface for changing the musical decisions before rendering a new recording. It is the strongest choice when the relationship between vocal melody and harmony matters more than getting a quick unplanned take.
Rank #3
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
Melody planning: useful for covers
Melody planning preserves a supplied melodic direction while allowing the accompaniment to be freer. The documented cover workflow uses an audio-to-score transcription model called SheetSage2 to obtain a melody score from a source recording. Check that transcription manually, align your new lyrics to the source sections, and then generate with cot="melody".
Planning off: a deliberate shortcut
cot="off" skips symbolic planning. Use it when a direct generation is more useful than an editable score, or as a controlled experiment against a planned version. Because there is no score to edit, changes generally require another generation from the input prompt and lyrics.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →ABC input: when you already have a score
Supported planning modes can accept ABC notation. This is useful when you have already authored or corrected a melody and chord progression. Validate the notation first; an incorrect score gives the model a poor musical constraint even when the lyrics and style prompt are excellent.
What is the exact WanGP workflow for generating a song with vocals?
Work from a clean project directory and change one variable at a time. The following sequence keeps the prompt, plan, render, and evaluation steps distinct.
- Create a project folder. Use a fresh output directory for this song version so files from earlier prompts cannot be mistaken for the current render.
- Write the style prompt. Include language, genre, BPM, vocal character, instruments, and the intended arrangement or mix character.
- Paste sectioned lyrics. Use explicit tags and check line lengths, spelling, and repeated hooks.
- Select planning. Start with
cot="full"for an original track. Supply ABC input only if you want to constrain the composition with an existing score. - Run generation. Let WanGP complete planning, semantic generation, synthesis, and decoding. Avoid judging a partially written file as the final result.
- Listen on neutral playback. Headphones are optional, but they can reveal vocal artifacts, clipped consonants, or masking that small speakers hide.
- Inspect saved metadata. Confirm the model identities, VAE, runtime, sampling information, seed, timings, and any truncation flag are present.
- Make one targeted change. Edit the score for melody or chords, revise a lyric section, or adjust the style prompt—but keep a copy of the previous artifact directory.
- Render to a new directory. Never overwrite the earlier version when comparing results.
The staged API underlying the workflow is plan() → generate_semantic() → synthesize() → decode(). You may encounter these stages in logs or advanced interfaces even when WanGP presents them as one button.
Can I edit the melody and chords after generation?
Yes, when you use a planning mode that produces a score. Edit the saved score, render it again, and treat the render as a new complete recording; the original waveform is not preserved inside the edited result.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWith full planning, the saved score.abc file is the principal musical control surface. Correct a phrase, alter a chord movement, or adjust the melody, then synthesize again. Keep the original audio and score in a separate directory so you can A/B the change and return to the unedited version.
Editing the score does not perform surgical waveform editing. A new render can change accompaniment, vocal phrasing, timing, or timbre beyond the exact notes you touched. For a production decision, compare the whole new recording rather than assuming only the edited bar changed.
Rank #4
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Broad Compatible: The usb c to usb adapter widely supports all USB C/ USB A devices like laptops, tablets, cellphones, car chargers, and phone chargers. Such as compatible with MacBook Pro/Air 2023/2022, Thunderbolt 4/3 Devices,Apple MagSafe Watch 9/8/7/SE/Ultra, iPad Pro 2022/2021, Samsung Galaxy S23/S20/S10, and iPhone 17/16/15 Pro. Plug and play
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
What files should I save for reproducible YuE2 results?
Save the complete artifact directory for every meaningful version. A playable FLAC alone is insufficient when you need to explain a difference, reproduce a result, or continue editing.
| Artifact | Why keep it |
|---|---|
audio.flac |
The rendered listening file. |
score.abc (when applicable) |
Editable melody and chord representation. |
plan.json |
Symbolic planning decisions and structure. |
| Semantic tokens | Intermediate representation used before acoustic synthesis. |
| Acoustic latents | Intermediate audio representation useful for diagnosis and comparison. |
| Effective configuration | The settings actually used, rather than the defaults you intended. |
| Timings | Stage durations that help identify slow or stalled steps. |
| Model identities | Exact model and component versions used for the render. |
| Integrity records | Evidence that files belong to the same generated artifact. |
Use a new directory after changing a prompt or score. Inspect truncation flags: a file can play normally while still ending early because a token limit was reached. Compare only outputs made with matching model, VAE, runtime, sampling, and seed information when you are evaluating a change.
Which decoder should I use for listening and benchmarking?
The normal listening path uses YuE2-Vae. YuE2-Vae-legacy is documented for reproducing the benchmark protocol, so keep benchmark-style outputs separate from everyday listening renders.
Do not mix decoder changes with prompt or score changes in the same comparison. If the decoder, model, runtime, and seed all change together, you cannot tell which factor caused a difference in vocal quality, timing, or noise.
How do I make a cover with YuE2‑3B?
A cover requires a melody score, lyric alignment, and rights review; it is not simply an original-song prompt with a familiar title. The documented workflow uses SheetSage2 to transcribe a source recording into a score, then uses YuE2 melody planning to build a new arrangement around that melodic guide.
- Obtain a lawful source. Make sure you have the right to analyze and use the recording for your intended purpose.
- Transcribe the melody. Use the identified audio-to-score workflow, then inspect the resulting score for wrong pitches, timing, rests, and repeated sections.
- Align lyrics. Map your licensed lyrics to the same verse, chorus, and bridge structure represented in the score.
- Set a target style. You can request a different genre, instrumentation, vocal character, language, and tempo while retaining the melodic guide.
- Generate with
cot="melody". Compare the output against the corrected score rather than trusting an unreviewed transcription. - Review publication rights. A generated vocal does not automatically clear the source recording, composition, or lyrics for commercial release.
Why is my YuE2 output cut short or missing sections?
First inspect the saved artifact and its truncation flags. A playable file can still have reached a token limit, so listening alone may not reveal why the final chorus or ending is absent.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Truncation flag is present: Treat the render as incomplete. Shorten the lyric structure, reduce unnecessary repetition, or adjust the relevant length setting available in your WanGP workflow, then render to a new directory.
- No truncation but a section is absent: Check the section tags and lyric text for malformed labels, then regenerate with a simpler structure.
- Cover ending is wrong: Inspect the transcribed score and section alignment. A melody transcription that drops a pickup or repeats a bar can make the generated arrangement appear to skip lyrics.
- Different versions are hard to compare: Preserve model, VAE, runtime, sampling, seed, and effective configuration in each artifact directory.
What should I do when WanGP cannot start YuE2?
Separate installation failures from generation failures. A model that does not appear in the interface points to files or discovery; a model that appears but crashes during generation points more often to runtime, memory, or kernel compatibility.
- YuE2 is not listed: Verify the expected YuE2 checkpoint layout, complete downloads, and restart WanGP.
- Python, PyTorch, CUDA, Triton, or attention errors: Follow the WanGP guidance for your installed GPU generation instead of mixing packages from an unrelated setup.
- Out-of-memory failure: Check whether a lower-memory or quantized option is available for your configuration, close competing GPU applications, and confirm that your card meets the documented 24GB BF16 baseline.
- First launch appears frozen: Check disk activity and network access; checkpoint and optional-asset downloads can be large and may take time.
- Audio is produced but sounds unstable: Simplify the style prompt, shorten or rebalance lyric lines, and test full planning before changing several runtime settings.
Can I publish or monetize songs made with YuE2‑3B?
Check the licenses separately for the repository code and the model weights before commercial distribution, client work, or monetized publishing. The YuE project states that its current code, agent skill, and documentation use Apache 2.0, while the model weights use CC BY-NC 4.0; WanGP’s YuE2 checkpoint documentation says its converted weights retain CC BY-NC 4.0.
Best Value
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
Those are different permissions. Apache 2.0 does not turn non-commercial model weights into commercial assets, and generating a song does not remove rights held in lyrics, source recordings, melodies, or samples. Read the current license text and obtain permission where required before releasing a paid product, advertising asset, client deliverable, or monetized upload.
What is the most reliable first-generation recipe?
Use a 24GB BF16-capable NVIDIA GPU, a fresh WanGP output directory, a compact style prompt, sectioned lyrics, and cot="full". Let the checkpoints finish downloading, save every artifact, and change only one input between versions.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →- Install WanGP with the runtime recommended for your GPU.
- Select YuE2 and complete checkpoint downloads.
- Enter a prompt such as “English, cinematic synth-pop, 112 BPM, intimate female lead vocal, pulsing bass, analog pads, crisp electronic drums, wide chorus, polished modern mix.”
- Paste lyrics with clearly labeled verses, pre-chorus, chorus, and bridge.
- Generate with full planning.
- Check the audio, score, metadata, and truncation status.
- Edit the score or lyrics only after identifying the specific problem.
- Render each revision into its own artifact directory and retain the model and runtime identity.
Frequently Asked Questions
Does YuE2‑3B run on an Apple Silicon Mac or an AMD GPU?
The upstream supported starting point specifies a BF16-capable NVIDIA GPU with 24GB VRAM, so Apple Silicon and AMD support should not be assumed from the cross-platform WanGP installers alone. Check the current WanGP hardware notes for your exact device before downloading checkpoints or planning a production workflow.
How long should my lyrics be for one YuE2 generation?
There is no universal lyric-length guarantee in the documented setup because output length depends on token limits and workflow settings. Start with one verse, a pre-chorus, a chorus, a second verse, a bridge, and a final chorus, then use the saved truncation indicators to decide whether the structure needs shortening.
Can I use a real singer’s voice in the style prompt?
Describe vocal qualities such as range, intimacy, breathiness, or intensity instead of requesting an identifiable living singer. A style prompt guides musical characteristics; it does not grant permission to imitate a person or use that performer’s identity commercially.
What does an ABC score file let me do outside WanGP?
ABC is a text-based representation of musical notation that can be inspected and edited before another render when the workflow supports score input. Keep a copy of the original file because a new render may change more than the notes you altered.
Should I keep FLAC or convert the song to another format?
Keep the generated FLAC as the archival master and make delivery copies separately if a platform requires another format. Converting the only copy and discarding the artifact directory removes useful evidence about the score, settings, model identity, and truncation status.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
