Upload folder using huggingface_hub

.github/workflows/deploy_to_hf.yml

@@ -21,8 +21,8 @@ jobs:
 
     - name: Install Poetry
       run: |
-        curl -sSL https://install.python-poetry.org | python3 -
-        echo "${{ runner.tool_cache }}/poetry/bin" >> $GITHUB_PATH
+        curl -sSL https://install.python-poetry.org | POETRY_VERSION=1.7.1 python3 -
+        echo "$HOME/.local/bin" >> $GITHUB_PATH
 
     - name: Export requirements.txt
       run: poetry export -f requirements.txt --output requirements.txt --without-hashes

README.md

@@ -12,7 +12,7 @@ license: mit
 ---
 # Improvisation Lab
 
-A Python package for generating musical improvisation melodies based on music theory principles. The package specializes in creating natural-sounding melodic phrases that follow chord progressions while respecting musical rules, with real-time pitch detection for practice feedback.
+A Python package for practicing musical improvisation through exercises. This package allows users to generate and practice melodic phrases based on music theory principles, offering real-time pitch detection for immediate feedback. Whether you're following chord progressions or practicing intervals, Improvisation Lab helps you improve your musical skills while adhering to musical rules.
 
 ## Try it out! 🚀
 <a href="https://huggingface.co/spaces/atsushieee/improvisation-lab" target="_blank">
@@ -21,16 +21,25 @@ A Python package for generating musical improvisation melodies based on music th
 
 Watch the demo in action:
 
+### Interval Practice: Demo
+
+https://github.com/user-attachments/assets/6a475cf0-9a82-4103-8316-7d4485b07c2e
+
+### Piece Practice: Demo
+
 https://github.com/user-attachments/assets/fa6e11d6-7b88-4b77-aa6e-a67c0927353d
 
 Experience Improvisation Lab directly in your browser! Our interactive demo lets you:
 
-- Generate melodic phrases based on chord progressions
+- Generate melodic phrases based on chord progressions or intervals
 - Practice your pitch accuracy in real-time
 - Get instant visual guidance for hitting the right notes
 
-Note: The demo runs on Hugging Face Spaces' free tier, which means:
+### Web Interface Features
+- **Tab Switching**: Easily switch between Interval Practice and Piece Practice using tabs in the web interface. This allows you to seamlessly transition between different practice modes without leaving the page.
 
+### Note
+The demo runs on Hugging Face Spaces' free tier, which means:
 - Performance might vary depending on server availability
 - If you encounter any issues, try refreshing the page or coming back later
 - For consistent performance, consider running the package locally
@@ -38,7 +47,17 @@ Note: The demo runs on Hugging Face Spaces' free tier, which means:
 
 ## Features
 
-- Generate melodic phrases based on scales and chord progressions
+- Web-based and direct microphone input support
+- Real-time pitch detection with FCPE (Fast Context-aware Pitch Estimation)
+- Provides real-time feedback on pitch accuracy
+
+### Interval Practice: Features
+- Focuses on practicing musical intervals.
+- Users can select the interval and direction (up or down) to practice.
+
+### Piece Practice: Features
+- Allows users to select a song and practice its chord progressions.
+- Generate melodic phrases based on scales and chord progressions.
 - Support for multiple scale types:
   - Major
   - Natural minor
@@ -57,6 +76,7 @@ Note: The demo runs on Hugging Face Spaces' free tier, which means:
 - Real-time pitch detection with FCPE (Fast Context-aware Pitch Estimation)
 - Web-based and direct microphone input support
 
+
 ## Prerequisites
 
 - Python 3.11 or higher
@@ -97,7 +117,7 @@ poetry run python main.py --app_type console
 
 The application can be customized through `config.yml` with the following options:
 
-#### Audio Settings
+#### Common Audio Settings
 - `sample_rate`: Audio sampling rate (default: 44100 Hz)
 - `buffer_duration`: Duration of audio processing buffer (default: 0.2 seconds)
 - `note_duration`: How long to display each note during practice (default: 3 seconds)
@@ -108,7 +128,12 @@ The application can be customized through `config.yml` with the following option
   - `f0_max`: Maximum frequency for the pitch detection algorithm (default: 880 Hz)
   - `device`: Device to use for the pitch detection algorithm (default: "cpu")
 
-#### Song Selection
+#### Interval Practice Settings
+- `interval`: The interval to practice
+  - Example: For a minor second descending interval, the interval value is -1
+- `num_problems`: The number of problems to practice
+
+#### Piece Practice Settings
 - `selected_song`: Name of the song to practice
 - `chord_progressions`: Dictionary of songs and their progressions
   - Format: `[scale_root, scale_type, chord_root, chord_type, duration]`
@@ -123,13 +148,20 @@ The application can be customized through `config.yml` with the following option
 
 ## How It Works
 
-### Melody Generation
+### Interval Practice: Melody Generation
+The interval practice focuses on improving interval recognition and singing accuracy:
+1. Users select the interval and direction (up or down) to practice.
+2. The application generates a series of problems based on the selected interval.
+3. Real-time feedback is provided to help users match the target interval.
+4. The practice session can be customized with the number of problems and note duration.
+
+### Piece Practice: Melody Generation
 The melody generation follows these principles:
-1. Notes are selected based on their relationship to the current chord and scale
-2. Chord tones have more freedom in movement
-3. Non-chord tones are restricted to moving to adjacent scale notes
-4. Phrases are connected naturally by considering the previous note
-5. All generated notes stay within the specified scale
+1. Notes are selected based on their relationship to the current chord and scale.
+2. Chord tones have more freedom in movement.
+3. Non-chord tones are restricted to moving to adjacent scale notes.
+4. Phrases are connected naturally by considering the previous note.
+5. All generated notes stay within the specified scale.
 
 ### Real-time Feedback
 Pitch Detection Demo:

improvisation_lab/application/interval_practice/web_interval_app.py

@@ -43,7 +43,7 @@ class WebIntervalPracticeApp(BasePracticeApp):
         self.results_table: List[List[Any]] = []
         self.progress_timer: float = 0.0
         self.is_auto_advance = False
-        self.note_duration = 1.5
+        self.note_duration = 3.0
 
     def _process_audio_callback(self, audio_data: np.ndarray):
         """Process incoming audio data and update the application state.
@@ -82,13 +82,11 @@ class WebIntervalPracticeApp(BasePracticeApp):
         self.update_results_table()
         self.current_note_idx += 1
         if self.current_note_idx >= len(self.phrases[self.current_phrase_idx]):
-            self.current_note_idx = 0
+            self.current_note_idx = 1
             self.current_phrase_idx += 1
             if self.current_phrase_idx >= len(self.phrases):
                 self.current_phrase_idx = 0
-            self.base_note = self.phrases[self.current_phrase_idx][
-                self.current_note_idx
-            ].value
+            self.base_note = self.phrases[self.current_phrase_idx][0].value
 
     def handle_audio(self, audio: Tuple[int, np.ndarray]) -> Tuple[str, str, str, List]:
         """Handle audio input from Gradio interface.
@@ -143,11 +141,10 @@ class WebIntervalPracticeApp(BasePracticeApp):
             num_notes=number_problems, interval=semitone_interval
         )
         self.current_phrase_idx = 0
-        self.current_note_idx = 0
+        self.current_note_idx = 1
         self.is_running = True
 
-        present_note = self.phrases[0][0].value
-        self.base_note = present_note
+        self.base_note = self.phrases[0][0].value
 
         if not self.audio_processor.is_recording:
             self.text_manager.initialize_text()
@@ -210,4 +207,4 @@ class WebIntervalPracticeApp(BasePracticeApp):
             result,
         ]
 
-        self.results_table.append(new_result)
+        self.results_table.insert(0, new_result)

improvisation_lab/domain/composition/melody_composer.py

@@ -87,5 +87,5 @@ class MelodyComposer:
         melody = []
         for base_note in base_notes:
             target_note = self.note_transposer.transpose_note(base_note, interval)
-            melody.append([base_note, target_note, base_note])
+            melody.append([base_note, target_note])
         return melody

improvisation_lab/presentation/interval_practice/web_interval_view.py

@@ -87,7 +87,7 @@ class WebIntervalPracticeView(WebPracticeView):
                 )
                 self.note_duration_box = gr.Number(
                     label="Note Duration (seconds)",
-                    value=1.5,
+                    value=3.0,
                 )
 
             self.generate_melody_button = gr.Button("Generate Melody")
@@ -97,6 +97,15 @@ class WebIntervalPracticeView(WebPracticeView):
             with gr.Row():
                 self.phrase_info_box = gr.Textbox(label="Problem Information", value="")
                 self.pitch_result_box = gr.Textbox(label="Pitch Result", value="")
+            """Create the audio input section."""
+            self.audio_input = gr.Audio(
+                label="Audio Input",
+                sources=["microphone"],
+                streaming=True,
+                type="numpy",
+                show_label=True,
+            )
+            self.end_practice_button = gr.Button("End Practice")
             self.results_table = gr.DataFrame(
                 headers=[
                     "Problem Number",
@@ -110,9 +119,7 @@ class WebIntervalPracticeView(WebPracticeView):
                 label="Result History",
             )
 
-            self._add_audio_input()
-            self.end_practice_button = gr.Button("End Practice")
-
+            self._add_audio_callbacks()
             self._add_buttons_callbacks()
 
             # Add Tone.js script
@@ -186,21 +193,13 @@ class WebIntervalPracticeView(WebPracticeView):
             outputs=[self.base_note_box, self.phrase_info_box, self.pitch_result_box],
         )
 
-    def _add_audio_input(self):
+    def _add_audio_callbacks(self):
         """Create the audio input section."""
-        audio_input = gr.Audio(
-            label="Audio Input",
-            sources=["microphone"],
-            streaming=True,
-            type="numpy",
-            show_label=True,
-        )
-
         # Attention: have to specify inputs explicitly,
         # otherwise the callback function is not called
-        audio_input.stream(
+        self.audio_input.stream(
             fn=self.on_audio_input,
-            inputs=audio_input,
+            inputs=self.audio_input,
             outputs=[
                 self.base_note_box,
                 self.phrase_info_box,

tests/domain/composition/test_melody_composer.py

@@ -94,7 +94,7 @@ class TestMelodyComposer:
 
         # Check the length of the melody
         assert len(melody) == len(base_notes)
-        assert len(melody[0]) == 3
+        assert len(melody[0]) == 2
 
         # Check the structure of the melody
         for i, base_note in enumerate(base_notes):
@@ -103,4 +103,3 @@ class TestMelodyComposer:
                 base_note, interval
             )
             assert melody[i][1] == transposed_note
-            assert melody[i][2] == base_note

tests/service/test_interval_practice_service.py

@@ -13,9 +13,9 @@ class TestPiecePracticeService:
         config = Config()
         service = IntervalPracticeService(config)
         melody = service.generate_melody(num_notes=10, interval=2)
-        # 10 notes, each with 3 parts (base, transposed, base)
+        # 10 notes, each with 2 parts (base, transposed)
         assert len(melody) == 10
-        assert all(len(note_group) == 3 for note_group in melody)
+        assert all(len(note_group) == 2 for note_group in melody)
         assert all(
             isinstance(note, Notes) for note_group in melody for note in note_group
         )