Mindket

Hum to Song

Newversion 1 · updated

Overview

Turns a 10 to 30 second hummed melody, a style line and lyrics into a produced song that keeps your tune. It suits songwriters and hobbyists who want to hear a hum developed into a track.

Hum to Song is a kit in the Writing category, listed by Mindket. It is free. This is version 1, last updated on . Download it on this page, a zip of its files. Nothing in it runs by itself.

At a glance

You need
  • Python 3.12
  • Python packages from requirements.txt, with Gradio 6.17.3
  • At Hugging Face the Space runs on zero-a10g hardware
Runs in
  • A web browser, on the Space's page at Hugging Face
  • Python 3.12 on your own Windows, macOS or Linux computer, using a downloaded copy
  • The listing carries an MCP tag, but the source gives no steps for using it that way
Start with
pip install -U huggingface_hub
hf download Mothersuperior/yue2-hum-to-song --repo-type space --local-dir yue2-hum-to-song
cd yue2-hum-to-song

Platform

WindowsOperating system
macOSOperating system
LinuxOperating system

About

What it is

Hum to Song is a Gradio app that turns a hummed melody into a produced song. You record a short hum of 10 to 30 seconds and type a style line and lyrics. A song comes back that keeps your tune. The song builds new sections around the hum and carries on after it stops.

Who it suits

People who have a tune in their head but no instruments or production skills to turn it into a finished track.

IndustryWho uses itHow they use itExample
MusicSongwritersSketch a song idea from a hummed melodyA chorus hummed into a phone
Hobby creationCasual makersTurn a tune and some lyrics into a full songA birthday song with custom words
EducationMusic teachersShow how a short motif grows into a structureA student's hummed phrase
ResearchAudio researchersTry a hum-conditioned song modelComparing the three melody modes

If you need music for commercial use, skip it, since the weights allow non-commercial use only.

Where to use it

  • A web browser, on the Space's page at Hugging Face
  • Python 3.12 on your own Windows, macOS or Linux computer, using a downloaded copy
  • The listing carries an MCP tag, but the source gives no steps for using it that way
Use caseWhoWhat they get back
Continue from humSongwritersA song that keeps the hummed melody and goes on from it
Hum onlyAnyoneA song that is exactly the hummed melody
Ignore hum melodyExperimentersA new tune where the hum only shapes the phrasing

How to set it up

  • Python 3.12
  • Python packages from requirements.txt, with Gradio 6.17.3
  • At Hugging Face the Space runs on zero-a10g hardware
  1. Install the Hugging Face hub package.
  2. Download the Space with the hf command.
  3. Move into the downloaded folder.
  4. Install Gradio 6.17.3, then the requirements.
  5. Start app.py.
pip install -U huggingface_hub
hf download Mothersuperior/yue2-hum-to-song --repo-type space --local-dir yue2-hum-to-song
cd yue2-hum-to-song
pip install gradio==6.17.3
pip install -r requirements.txt
python app.py

How to use it

Hum a melody for 10 to 30 seconds, then add a style line and lyrics and run it. Pick one of three melody modes. Max song length is only a cap, since the model decides where the song ends.

python app.py

Example

You hum a short tune, type a style line such as a slow acoustic ballad, and add two lines of lyrics. The app returns a produced song that follows your melody and keeps going after the hum ends.

For newcomers and experts

Start online. Open the Space page, hum a real melody into your microphone for about 15 seconds, add a style and lyrics, and listen to the result. A real hum works, but a synthesised tone does not.

The app pairs the Mothersuperior/YuE2-hum-to-song adapter with the m-a-p/YuE2-3B model. SheetSage2 transcribes the hum into a score, which is left open so the planner keeps writing. A prosody adapter then feeds a pitch carrier made from the hum into the decoder. The adapter is subtle, so judge it over the first hummed section.

  1. You hum 10 to 30 seconds
  2. SheetSage2 turns the hum into a score
  3. The planner writes the rest of the song
  4. The decoder renders the audio

Good to know

  • Your hum decides the tune and how the song starts, while the model decides how long it runs
  • The transcriber needs something voice-like, so a synthesised tone fails
  • Hardware: zero-a10g at Hugging Face, with Gradio 6.17.3 and Python 3.12
  • License: cc-by-nc-4, because the weights derive from YuE2-3B, so non-commercial use only
  • Cost to run: none named
  • Data it reaches: your hum, style line and lyrics, plus the models and the example hum dataset it names
  • Not for: commercial use

README and files

README

Hum to Song

Turns a 10 to 30 second hummed melody, a style line and lyrics into a produced song that keeps your tune. It suits songwriters and hobbyists who want to hear a hum developed into a track.

Made by Mothersuperior on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: Hum to Song is a Gradio app that turns a hummed melody into a produced song. You record a short hum of 10 to 30 seconds and type a style line and lyrics. A song comes back that keeps your tune. The song builds new sections around the hum and carries on after it stops.

Who it suits: People who have a tune in their head but no instruments or production skills to turn it into a finished track.

Industry | Who uses it | How they use it | Example
Music | Songwriters | Sketch a song idea from a hummed melody | A chorus hummed into a phone
Hobby creation | Casual makers | Turn a tune and some lyrics into a full song | A birthday song with custom words
Education | Music teachers | Show how a short motif grows into a structure | A student's hummed phrase
Research | Audio researchers | Try a hum-conditioned song model | Comparing the three melody modes

If you need music for commercial use, skip it, since the weights allow non-commercial use only.

Where to use it:

  • A web browser, on the Space's page at Hugging Face
  • Python 3.12 on your own Windows, macOS or Linux computer, using a downloaded copy
  • The listing carries an MCP tag, but the source gives no steps for using it that way
Use case | Who | What they get back
Continue from hum | Songwriters | A song that keeps the hummed melody and goes on from it
Hum only | Anyone | A song that is exactly the hummed melody
Ignore hum melody | Experimenters | A new tune where the hum only shapes the phrasing

How to set it up:

  • Python 3.12
  • Python packages from requirements.txt, with Gradio 6.17.3
  • At Hugging Face the Space runs on zero-a10g hardware
  1. Install the Hugging Face hub package.
  2. Download the Space with the hf command.
  3. Move into the downloaded folder.
  4. Install Gradio 6.17.3, then the requirements.
  5. Start app.py.
pip install -U huggingface_hub
hf download Mothersuperior/yue2-hum-to-song --repo-type space --local-dir yue2-hum-to-song
cd yue2-hum-to-song
pip install gradio==6.17.3
pip install -r requirements.txt
python app.py

How to use it: Hum a melody for 10 to 30 seconds, then add a style line and lyrics and run it. Pick one of three melody modes. Max song length is only a cap, since the model decides where the song ends.

python app.py

Example: You hum a short tune, type a style line such as a slow acoustic ballad, and add two lines of lyrics. The app returns a produced song that follows your melody and keeps going after the hum ends.

For newcomers and experts: Start online. Open the Space page, hum a real melody into your microphone for about 15 seconds, add a style and lyrics, and listen to the result. A real hum works, but a synthesised tone does not.

The app pairs the Mothersuperior/YuE2-hum-to-song adapter with the m-a-p/YuE2-3B model. SheetSage2 transcribes the hum into a score, which is left open so the planner keeps writing. A prosody adapter then feeds a pitch carrier made from the hum into the decoder. The adapter is subtle, so judge it over the first hummed section.

You hum 10 to 30 seconds
SheetSage2 turns the hum into a score
The planner writes the rest of the song
The decoder renders the audio

Good to know:

  • Your hum decides the tune and how the song starts, while the model decides how long it runs
  • The transcriber needs something voice-like, so a synthesised tone fails
  • Hardware: zero-a10g at Hugging Face, with Gradio 6.17.3 and Python 3.12
  • License: cc-by-nc-4, because the weights derive from YuE2-3B, so non-commercial use only
  • Cost to run: none named
  • Data it reaches: your hum, style line and lyrics, plus the models and the example hum dataset it names
  • Not for: commercial use

On Hugging Face, 13 November 2026

  • Likes: 124
  • Built with: gradio
  • License: cc-by-nc-4.0

Links

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.

Files
README.md  4 KB

1 file, 4 KB, downloaded as a zip.

What’s new

  • Version 1 · 2026-11-12

Compare with similar

Listed Hugging Face Spaces in the same category that share the most tags with this one.

ProductLikesBuilt withLicenseLast change
Hum to Song This page124gradiocc-by-nc-4.0
PROMPT++Shares gradio247gradioapache-2.0
humanizerShares gradio51gradioapache-2.0
Arabic Poetry GeneratorShares gradio30gradiocc-by-nc-4.0
Essay_generator_appShares gradio4gradioapache-2.0
Sakil_text_summarization_appShares gradio3gradioapache-2.0
Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13.

Reviews

No ratings yet

Write a review

No reviews yet.

More from Mindket

See all from Mindket

anaksor-ocr

Sokheng · Kit

Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…

New

Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…

New

ACE-Step 1.5 XL SFT

ACE-Step · Kit

Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…

New

Similar in Writing

See all in Writing

Rewrites English text with fewer grammar mistakes, using a T5 model trained on the JFLEG dataset. Made by vennify for writers, students and developers who want…

New

Translates between 33 languages on your own device with a 1.8B model squeezed to 1.25 bits, about 440 MB as GGUF files for llama.cpp. For developers adding off…

New

Summarizes, answers questions about and continues book-length text with a fine-tuned 27B model that uses Gated DeltaNet layers, in English and Japanese. For pe…

New