|Tunes||Improvisations||Bird Song||New Fractal Tunes||Recordings||Test my Midi player|
Send the birdsong of your choice as a musical e-card
Here are some transcriptions of bird song using my Fractal Tune Smithy program.
The original clips are from: European Birds, songs and sonograms.
Zip of them all: bird_song.zip.
Most are played on the Midi recorder voice.
Some are for oboe, or Shanai (a kind of shawm) for the Golden Eagle.
For thrushes, used flute and oboe in unison. For Red Kite, used high bassoon and violin in unison.
Here is a fractal tune for the curlew song:
Tune Smithy file - this uses a new option for the fractal tunes so you won't hear it exactly as in the mid clip until next beta release:
Tune Smithy files for all the midi clips: bird_song_ts.zip
If one did transcriptions by ear (using slowed down bird song) one would find more notes, but it would take quite a while.
For the automatic transcriptions, you set a minimum note length for FTS to search for and tweak various other parameters until it is finding as many of the notes as one can get it to find. If a note glissandos or has vibrato, FTS will find the average pitch, or a series of pitches for a long glissando (if not too fast)
It's done by counting note crossings. This seems to be much more accurate than Fast Fourier Transforms when it is appropriate.
You can get accuracy of about +-0.1 cents for a 0.1 second note if note is optimal, such as sine wave (by linear interpolation between the two samples before and after change of sign). Some birdsong is pretty close to sinewaves. For other ones, you can set a height of the secondary peaks to ignore in FTS. This also works fine for recorder, and also for singing for some voice timbres (can depend on which vowel sound you sing).
Here are two midi clips transcribed from Veena Sahasrabuddhe's site - her voice works well.
Veena_Sahasrabuddhe_phrase1.mid, and Veena_Sahasrabuddhe_phrase2.mid
(just for demonstration purposes!)
It's from: http://stations.mp3s.com/stations/123/santhaa_notes.html Alap1 where she sings phrases slowly and clearly for beginner singers to learn, so ideal for my program.
Adds extra notes at times - e.g. from the attack - means there was enough of that pitch in the attack for the program to recognise it as an apparently new note.
In a glissando, my program follows the gliss, and adds new notes every so often.
Also works well for some soundcard voices, and is useful for checking the relative and absolute pitch accuracy of your soundcard pitch bends.
Anyway, this isn't ready yet, but is a taste of what you will be able to do with the next FTS beta when ready.