| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

Text2Song

Page history last edited by eliram 15 years, 10 months ago
[דרור גיל]
כחובב מוסיקה נלהב, בכל פעם שמישהו אומר מילה, ביטוי או משפט שלם, תמיד זה מזכיר לי איזשהו שיר שמתחיל להתנגן לי בראש.
אז לכבוד מיוזנט, חשבתי למחשב את התהליך:  ליצור תוכנה שלוקחת משפט כתוב, ומשמיעה אותו מקטעי שירים.  התשתמש מקליד בתוכנה משפט או ביטוי, והתוכנה בונה את המשפט מקטעי אודיו קצרים (מילים) שנחתכו משירים, ומשמיעה את המשפט.  התוכנה גם תספק מידע על השירים ששימשו לבניית המשפט.
 
לדוגמא, אם המשתמש מקליד את המשפט:
I like to meet cool people
 
התוכנה תשמיע את המשפט מקטעי אודיו של מילות שירים, ותציג את המידע הבא:

 

I Like to Meet Cool People

"I" taken from the song "I Love You Always Forever" by Donna Lewis

"Like" taken from the song "I Don't Like Mondays" by The Boomtown Rats

"To" taken from the song "Nine to Five" by Dolly Parton

"Meet" taken from the song "I'll Meet You at Midnight" by Smokie

"Cool" taken from the song "Smooth" by Santana

"People" taken form the song "People are People" by Depeche Mode

 

אם יש מילה במשפט שלא נמצאת במאגר המילים של התוכנה, התוכנה תשמיע אותה באמצעות סנתז דיבור (Text to Speech).
 
אלירם התנדב לכתוב את התוכנה, וגם הפנה את תשומת לבי לאתר אינטרנט שכבר עושה את זה.  אבל אנחנו נבנה מערכת שעובדת גם בעברית. וגם באנגלית.
אני התחלתי להכין אוסף של קבצי אודיו קצרים בפורמט MP3 כששם הקובץ הוא שם המילה שבתוכו (למשל meet. mp3 או mp3 .עולם).  מכיוון שאיכות הפרויקט תשתפר ככל שיהיו יותר מילים, נשמח עם  עוד אנשים יצטרפו למאמץ של הכנת קבצי המילים.  ניתן להיעזר בכל תוכנות אודיו, כמו למשל Audacity או Goldwave.  מי שמוכן לעזור, שיפנה אלי  באימייל (הג'ימייל שלי הוא drorgill).  בנוסף, אנחנו נקליט אנשים ששרים עוד מילים בזמן MuseNet, ונוסיף אותם למאגר המילים במקום.
 
[אלירם חקלאי]
הרעיון של דרור היה מעולה וכדרכם של רעיונות מעולים, גרסה שלהם כבר נוצרה במקום אחר. מכיוון שכך, במקום לחזור על דברים שנעשו פיתחנו את הרעיון. חוץ מלהוסיף תמיכה במלים בעברית (ולמעשה בכל שפה, המגבלה היא הסימפולים הקיימים) הוספתי אלמנטים של משחק במלים בווריאציה משוכללת על הלוגו המנגן של מיוזנט. המלים תופסות חיים משל עצמן על המסך, אפשר לשחק אתן והן גם מתרבות בהתאם.

 

[Dror Gill]

As an avid music lover, everytime someone says a word, phrase or full sentence, it always reminds me of a song that immediately starts playing in my head.

So I thought of automating this process as a project for MuseNet: Create software that takes a phrase of text, and constructs it from song pieces.  The user types in a phrase or a sentence, and the software builds this sentence from short audio pieces (words) which are cut from songs, and plays the result.  The software would also supply some information about the songs used to build the sentence.

 

For example, if the user types the sentence:

I like to meet cool people

 

The software would play this sentence from audio pieces (words) taken from songs, and would display the following information:

 

I Like to Meet Cool People

"I" taken from the song "I Love You Always Forever" by Donna Lewis

"Like" taken from the song "I Don't Like Mondays" by The Boomtown Rats

"To" taken from the song "Nine to Five" by Dolly Parton

"Meet" taken from the song "I'll Meet You at Midnight" by Smokie

"Cool" taken from the song "Smooth" by Santana

"People" taken form the song "People are People" by Depeche Mode

 

If the software can't find a certain word in its database, it would play it out using a speech synthesizer (text to speech).

 

Eliram volunteered to write the software, and also pointed me to an existing website that already does this.  But we will do it even better - both in Hebrew and in English.

I started to prepare short audio files in

MP3

format, where the name of the file will be the same as the word contained in the file (for example meet.

mp3

).  Since the quality of the project will improve as we have more words, we are looking for volunteers to help in creating the word files.  You can use any audio software, such as Audacity or Goldwave.  If you are willing to help, please send me email to my gmail which is drorgill.  We will also record people singing some words at MuseNet, and add them to the database on the spot.

 

[Eliram Haklai]

This was a great idea by Dror, but as often happens with great ideas, someone already created some version of it. So we decided to further develop the idea and aside from adding support to Hebrew (or any other language, just add the right samples) I've added some game elements and the ability to play with the words rather than just play them. They interact with the player and multiply accordingly.

Comments (4)

eliram said

at 10:46 pm on May 24, 2008

I recall visiting a site that did something similar a few years ago. It has samples from many English songs and you could enter a sentence and listen to the words taken from different songs.
I can't find the site now, but I'll keep looking.

It shouldn't be too hard to do this with Flash, I'll be happy to help.

eliram said

at 3:44 pm on May 25, 2008

Found the website. I knew it was in the internet somewhere:
http://www.sr.se/p1/src/sing/

They did a great job, it could be nice to display that site in MuseNet and let people write sentences.

Dror Gill said

at 8:03 am on May 26, 2008

Cool stuff, but let's do our own...

eliram said

at 2:00 pm on May 26, 2008

You know what? It's MuseNet, right?
Let's do this LIVE. Let's allow people to sing the words from the songs THEMSELVES, record them and add their words into the repository.

This will:
1. Allow people to participate and donate live
2. Eliminate copyright issues when using original recordings, so we can later on upload this to the site.
3. Let us build a database during MuseNet and at the end see how many words we could record in a day (And which words).
4. We could use both Hebrew and English words and songs.
5. We can record the speeches at the beginning of the day and extract words from the speeches as well.

Requirements:
- Computer
- Microphone
- Editing software (Audacity)
- Text Editor (I rather use a XML file and not read ID3 tags).

You don't have permission to comment on this page.