lyrics-transcriber 0.12.8__tar.gz → 0.13.1__tar.gz

{lyrics_transcriber-0.12.8 → lyrics_transcriber-0.13.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: lyrics-transcriber
-Version: 0.12.8
+Version: 0.13.1
 Summary: Automatically create synchronised lyrics files in ASS and MidiCo LRC formats with word-level timestamps, using Whisper and lyrics from Genius and Spotify
 Home-page: https://github.com/karaokenerds/python-lyrics-transcriber
 License: MIT
@@ -13,21 +13,22 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
-Requires-Dist: Cython (>=0,<1)
-Requires-Dist: dtw-python (>=1,<2)
-Requires-Dist: llvmlite (>=0,<1)
-Requires-Dist: lyricsgenius (>=3,<4)
-Requires-Dist: numba (>=0.57,<0.58)
-Requires-Dist: numpy (>=1,<2)
-Requires-Dist: onnx (>=1,<2)
-Requires-Dist: onnxruntime (>=1,<2)
+Requires-Dist: Cython (>=0)
+Requires-Dist: dtw-python (>=1)
+Requires-Dist: llvmlite (>=0)
+Requires-Dist: lyricsgenius (>=3)
+Requires-Dist: numba (>=0.57)
+Requires-Dist: numpy (>=1)
+Requires-Dist: onnx (>=1)
+Requires-Dist: onnxruntime (>=1)
 Requires-Dist: openai (>=1,<2)
-Requires-Dist: openai-whisper (==20231117)
-Requires-Dist: python-slugify (>=8,<9)
-Requires-Dist: syrics (>=0,<1)
-Requires-Dist: torch (>1)
-Requires-Dist: tqdm (>=4,<5)
-Requires-Dist: whisper-timestamped (>=1,<2)
+Requires-Dist: openai-whisper (>=20231117)
+Requires-Dist: python-slugify (>=8)
+Requires-Dist: syrics (>=0)
+Requires-Dist: torch (>=1)
+Requires-Dist: tqdm (>=4)
+Requires-Dist: transformers (>=4)
+Requires-Dist: whisper-timestamped (>=1)
 Project-URL: Documentation, https://github.com/karaokenerds/python-lyrics-transcriber/blob/main/README.md
 Project-URL: Repository, https://github.com/karaokenerds/python-lyrics-transcriber
 Description-Content-Type: text/markdown

lyrics_transcriber-0.13.1/lyrics_transcriber/llm_prompts/promptfooconfig.yaml

@@ -0,0 +1,61 @@
+# This configuration runs each prompt through a series of example inputs and checks if they meet requirements.
+# Learn more: https://promptfoo.dev/docs/configuration/guide
+
+description: Song lyric corrector for a karaoke video studio, responsible for reading lyrics inputs, correcting them and generating JSON-based responses containing the corrected lyrics according to predefined criteria.
+providers:
+  - id: openai:gpt-3.5-turbo-1106
+    config:
+      temperature: 0
+  # - id: openai:gpt-4-1106-preview
+  #   config:
+  #     temperature: 0
+prompts:
+  - file://llm_prompt_lyrics_correction_andrew_handwritten_20231118.txt
+
+defaultTest:
+  assert:
+    - type: is-json
+      value:
+        required: [id, text, words]
+        type: object
+        properties:
+          id:
+            type: number
+          text:
+            type: string
+          words:
+            type: array
+            items:
+              type: object
+              properties:
+                text:
+                  type: string
+                start:
+                  type: number
+                end:
+                  type: number
+                confidence:
+                  type: number
+
+tests:
+  - description: ABBA - Under Attack (segment 0)
+    vars:
+      reference_lyrics: file://test_data/ABBA-UnderAttack-Genius.txt
+      previous_two_corrected_lines: 
+      upcoming_two_uncorrected_lines: 
+      segment_input: |
+        {"id": 0, "start": 17.46, "end": 21.3, "confidence": 0.792, "text": " Don't know how to take it, don't know where to go", "words": [{"text": "Don't", "start": 17.46, "end": 18.2, "confidence": 0.278}, {"text": "know", "start": 18.2, "end": 18.42, "confidence": 0.965}, {"text": "how", "start": 18.42, "end": 18.66, "confidence": 0.865}, {"text": "to", "start": 18.66, "end": 18.88, "confidence": 0.994}, {"text": "take", "start": 18.88, "end": 19.2, "confidence": 0.992}, {"text": "it,", "start": 19.2, "end": 19.44, "confidence": 0.974}, {"text": "don't", "start": 19.56, "end": 19.8, "confidence": 0.917}, {"text": "know", "start": 19.8, "end": 20.02, "confidence": 0.989}, {"text": "where", "start": 20.02, "end": 20.46, "confidence": 0.963}, {"text": "to", "start": 20.46, "end": 20.76, "confidence": 0.983}, {"text": "go", "start": 20.76, "end": 21.3, "confidence": 0.982}]}
+    assert:
+      - type: contains
+        value: "Don't know how to take it, don't know where to go"
+
+  - description: ABBA - Under Attack (segment 1)
+    vars:
+      reference_lyrics: file://test_data/ABBA-UnderAttack-Genius.txt
+      previous_two_corrected_lines: 
+      upcoming_two_uncorrected_lines: 
+      segment_input: |
+        {"id": 1, "start": 22.04, "end": 27.84, "confidence": 0.763, "text": " My resistance running low And every day the hole is getting tighter", "words": [{"text": "My", "start": 22.04, "end": 22.32, "confidence": 0.535}, {"text": "resistance", "start": 22.32, "end": 22.94, "confidence": 0.936}, {"text": "running", "start": 22.94, "end": 23.66, "confidence": 0.89}, {"text": "low", "start": 23.66, "end": 24.36, "confidence": 0.999}, {"text": "And", "start": 24.36, "end": 25.14, "confidence": 0.485}, {"text": "every", "start": 25.14, "end": 25.56, "confidence": 0.568}, {"text": "day", "start": 25.56, "end": 25.88, "confidence": 0.997}, {"text": "the", "start": 25.88, "end": 26.1, "confidence": 0.959}, {"text": "hole", "start": 26.1, "end": 26.48, "confidence": 0.361}, {"text": "is", "start": 26.48, "end": 26.68, "confidence": 0.947}, {"text": "getting", "start": 26.68, "end": 27.08, "confidence": 0.996}, {"text": "tighter", "start": 27.08, "end": 27.84, "confidence": 0.975}]}
+    assert:
+      - type: contains
+        value: "My resistance running low And every day the hold is getting tighter"

lyrics_transcriber-0.13.1/lyrics_transcriber/llm_prompts/test_data/ABBA-UnderAttack-Genius.txt

@@ -0,0 +1,48 @@
+Don't know how to take it, don't know where to go
+My resistance running low
+And every day the hold is getting tighter and it troubles me so
+(You know that I'm nobody's fool)
+I'm nobody's fool and yet it's clear to me
+I don't have a strategy
+It's just like taking candy from a baby and I think I must be
+
+Under attack, I'm being taken
+About to crack, defences breaking
+Won't somebody please have a heart
+Come and rescue me now 'cause I'm falling apart
+Under attack, I'm taking cover
+He's on my track, my chasing lover
+Thinking nothing can stop him now
+Should I want to, I'm not sure I would know how
+
+This is getting crazy, I should tell him so
+Really let my anger show
+Persuade him that the answer to his questions is a definite no
+(I'm kind of flattered I suppose)
+Guess I'm kind of flattered but I'm scared as well
+Something like a magic spell
+I hardly dare to think of what would happen, where I'd be if I fell
+
+Under attack, I'm being taken
+About to crack, defences breaking
+Won't somebody please have a heart
+Come and rescue me now 'cause I'm falling apart
+Under attack, I'm taking cover
+He's on my track, my chasing lover
+Thinking nothing's gonna stop him now
+Should I want to, I'm not sure I won't know how
+
+Under attack, I'm being taken
+About to crack, defences breaking
+Won't somebody see and save a heart
+Come and rescue me now 'cause I'm falling apart
+Under attack, I'm taking cover
+He's on my track, my chasing lover
+Thinking nothing can stop him now
+Should I want to, I'm not sure I would know how
+
+Under attack, I'm being taken
+About to crack, defences breaking
+Won't somebody please have a heart
+Come and rescue me now 'cause I'm falling apart
+Under attack, I'm taking cover

{lyrics_transcriber-0.12.8 → lyrics_transcriber-0.13.1}/lyrics_transcriber/transcriber.py

@@ -190,6 +190,9 @@ class LyricsTranscriber:
             online_lyrics_text_key = f"{online_lyrics_source}_lyrics_text"
             online_lyrics_filepath_key = f"{online_lyrics_source}_lyrics_filepath"
 
+            if online_lyrics_text_key not in self.outputs:
+                continue
+
             data_input_str = (
                 f'Data input 1:\n{self.outputs["transcribed_lyrics_text"]}\nData input 2:\n{self.outputs[online_lyrics_text_key]}\n'
             )
@@ -274,7 +277,7 @@ class LyricsTranscriber:
         total_segments = len(self.outputs["transcription_data_dict"]["segments"])
         self.logger.info(f"Beginning correction using LLM, total segments: {total_segments}")
 
-        with open(self.outputs["llm_transcript_filepath"], "a", buffering=1) as llm_transcript_file:
+        with open(self.outputs["llm_transcript_filepath"], "a", buffering=1, encoding="utf-8") as llm_transcript_file:
             self.logger.debug(f"writing LLM chat instructions: {self.outputs['llm_transcript_filepath']}")
 
             llm_transcript_header = f"--- SYSTEM instructions passed in for all segments ---:\n\n{system_prompt}\n"
@@ -303,14 +306,13 @@ class LyricsTranscriber:
                 previous_two_corrected_lines = ""
                 upcoming_two_uncorrected_lines = ""
 
-                if segment["id"] > 2:
-                    for previous_segment in corrected_lyrics_dict["segments"]:
-                        if previous_segment["id"] in (segment["id"] - 2, segment["id"] - 1):
-                            previous_two_corrected_lines += previous_segment["text"].strip() + "\n"
+                for previous_segment in corrected_lyrics_dict["segments"]:
+                    if previous_segment["id"] in (segment["id"] - 2, segment["id"] - 1):
+                        previous_two_corrected_lines += previous_segment["text"].strip() + "\n"
 
-                    for next_segment in self.outputs["transcription_data_dict"]["segments"]:
-                        if next_segment["id"] in (segment["id"] + 1, segment["id"] + 2):
-                            upcoming_two_uncorrected_lines += next_segment["text"].strip() + "\n"
+                for next_segment in self.outputs["transcription_data_dict"]["segments"]:
+                    if next_segment["id"] in (segment["id"] + 1, segment["id"] + 2):
+                        upcoming_two_uncorrected_lines += next_segment["text"].strip() + "\n"
 
                 llm_transcript_segment += f"--- Segment {segment['id']} / {total_segments} ---\n"
                 llm_transcript_segment += f"Previous two corrected lines:\n\n{previous_two_corrected_lines}\nUpcoming two uncorrected lines:\n\n{upcoming_two_uncorrected_lines}\nData input:\n\n{segment_input}\n"
@@ -371,7 +373,7 @@ class LyricsTranscriber:
             self.logger.info(f'Successfully processed correction for all {len(corrected_lyrics_dict["segments"])} lyrics segments')
 
             self.logger.debug(f"writing corrected lyrics data JSON filepath: {corrected_lyrics_data_json_cache_filepath}")
-            with open(corrected_lyrics_data_json_cache_filepath, "w") as corrected_lyrics_data_json_cache_file:
+            with open(corrected_lyrics_data_json_cache_filepath, "w", encoding="utf-8") as corrected_lyrics_data_json_cache_file:
                 corrected_lyrics_data_json_cache_file.write(json.dumps(corrected_lyrics_dict, indent=4))
 
         self.outputs["corrected_lyrics_data_filepath"] = corrected_lyrics_data_json_cache_filepath
@@ -406,7 +408,7 @@ class LyricsTranscriber:
             self.outputs["corrected_lyrics_text"] = ""
 
             self.logger.debug(f"writing lyrics plain text to corrected_lyrics_text_filepath: {corrected_lyrics_text_filepath}")
-            with open(corrected_lyrics_text_filepath, "w") as f:
+            with open(corrected_lyrics_text_filepath, "w", encoding="utf-8") as f:
                 for corrected_segment in self.outputs["corrected_lyrics_data_dict"]["segments"]:
                     self.outputs["corrected_lyrics_text"] += corrected_segment["text"].strip() + "\n"
                     f.write(corrected_segment["text".strip()] + "\n")
@@ -453,7 +455,7 @@ class LyricsTranscriber:
             self.logger.debug(
                 f"writing lyrics data JSON to spotify_lyrics_data_json_cache_filepath: {spotify_lyrics_data_json_cache_filepath}"
             )
-            with open(spotify_lyrics_data_json_cache_filepath, "w") as f:
+            with open(spotify_lyrics_data_json_cache_filepath, "w", encoding="utf-8") as f:
                 f.write(spotify_lyrics_json)
         except Exception as e:
             self.logger.warn(f"caught exception while attempting to fetch from spotify: ", e)
@@ -473,7 +475,7 @@ class LyricsTranscriber:
             self.outputs["spotify_lyrics_text"] = ""
 
             self.logger.debug(f"writing lyrics plain text to spotify_lyrics_text_filepath: {spotify_lyrics_text_filepath}")
-            with open(spotify_lyrics_text_filepath, "w") as f:
+            with open(spotify_lyrics_text_filepath, "w", encoding="utf-8") as f:
                 for line in lines:
                     self.outputs["spotify_lyrics_text"] += line["words"].strip() + "\n"
                     f.write(line["words"].strip() + "\n")
@@ -505,7 +507,7 @@ class LyricsTranscriber:
         lyrics = self.clean_genius_lyrics(song.lyrics)
 
         self.logger.debug(f"writing clean lyrics to genius_lyrics_cache_filepath: {genius_lyrics_cache_filepath}")
-        with open(genius_lyrics_cache_filepath, "w") as f:
+        with open(genius_lyrics_cache_filepath, "w", encoding="utf-8") as f:
             f.write(lyrics)
 
         self.outputs["genius_lyrics_filepath"] = genius_lyrics_cache_filepath
@@ -559,7 +561,7 @@ class LyricsTranscriber:
 
         lrc_filename = self.outputs["midico_lrc_filepath"]
         self.logger.debug(f"writing midico formatted word timestamps to LRC file: {lrc_filename}")
-        with open(lrc_filename, "w") as f:
+        with open(lrc_filename, "w", encoding="utf-8") as f:
             f.write("[re:MidiCo]\n")
             for segment in self.outputs["corrected_lyrics_data_dict"]["segments"]:
                 for i, word in enumerate(segment["words"]):
@@ -753,7 +755,7 @@ class LyricsTranscriber:
             self.outputs["transcribed_lyrics_text"] = ""
 
             self.logger.debug(f"writing lyrics plain text to transcribed_lyrics_text_filepath: {transcribed_lyrics_text_filepath}")
-            with open(transcribed_lyrics_text_filepath, "w") as f:
+            with open(transcribed_lyrics_text_filepath, "w", encoding="utf-8") as f:
                 for segment in self.outputs["transcription_data_dict"]["segments"]:
                     self.outputs["transcribed_lyrics_text"] += segment["text"] + "\n"
                     f.write(segment["text"].strip() + "\n")

{lyrics_transcriber-0.12.8 → lyrics_transcriber-0.13.1}/lyrics_transcriber/utils/cli.py

@@ -2,7 +2,6 @@
 import argparse
 import logging
 import pkg_resources
-from lyrics_transcriber import LyricsTranscriber
 
 
 def main():
@@ -110,6 +109,9 @@ def main():
 
     logger.debug("Loading LyricsTranscriber class")
 
+    # Lazy load this class so help output is printed quickly rather than waiting for heavy libraries to load
+    from lyrics_transcriber import LyricsTranscriber
+
     transcriber = LyricsTranscriber(
         args.audio_filepath,
         genius_api_token=args.genius_api_token,

{lyrics_transcriber-0.12.8 → lyrics_transcriber-0.13.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "lyrics-transcriber"
-version = "0.12.8"
+version = "0.13.1"
 description = "Automatically create synchronised lyrics files in ASS and MidiCo LRC formats with word-level timestamps, using Whisper and lyrics from Genius and Spotify"
 authors = ["Andrew Beveridge <andrew@beveridge.uk>"]
 license = "MIT"
@@ -13,21 +13,22 @@ documentation = "https://github.com/karaokenerds/python-lyrics-transcriber/blob/
 
 [tool.poetry.dependencies]
 python = ">=3.9"
-Cython = "^0"
-dtw-python = "^1"
-llvmlite = "^0"
-numba = "^0.57"
-numpy = "^1"
-onnx = "^1"
-onnxruntime = "^1"
-torch = ">1"
-tqdm = "^4"
-lyricsgenius = "^3"
-python-slugify = "^8"
-syrics = "^0"
+Cython = ">=0"
+dtw-python = ">=1"
+llvmlite = ">=0"
+numba = ">=0.57"
+numpy = ">=1"
+onnx = ">=1"
+onnxruntime = ">=1"
+torch = ">=1"
+tqdm = ">=4"
+lyricsgenius = ">=3"
+python-slugify = ">=8"
+syrics = ">=0"
 openai = "^1"
-openai-whisper = "20231117"
-whisper-timestamped = "^1"
+openai-whisper = ">=20231117"
+transformers = ">=4"
+whisper-timestamped = ">=1"
 # Note: after adding openai-whisper and whisper-timestamped with poetry lock, I then removed all traces of triton
 # from poetry.lock before running poetry install, as triton doesn't support macOS but isn't actually needed for whisper.
 # This was the only way I was able to get a working cross-platform build published to PyPI.

lyrics_transcriber-0.12.8/lyrics_transcriber/llm_prompts/promptfooconfig.yaml

@@ -1,39 +0,0 @@
-# This configuration runs each prompt through a series of example inputs and checks if they meet requirements.
-# Learn more: https://promptfoo.dev/docs/configuration/guide
-
-prompts:
-  - file://llm_prompt_lyrics_correction_*.txt
-providers: [openai:gpt-3.5-turbo-0613, openai:gpt-4-1106-preview]
-tests:
-  - description: First test case - automatic review
-    vars:
-      var1: first variable's value
-      var2: another value
-      var3: some other value
-    # For more information on assertions, see https://promptfoo.dev/docs/configuration/expected-outputs
-    assert:
-      - type: equals
-        value: expected LLM output goes here
-      - type: contains
-        value: some text
-      - type: javascript
-        value: 1 / (output.length + 1)  # prefer shorter outputs
-
-  - description: Second test case - manual review
-    # Test cases don't need assertions if you prefer to manually review the output
-    vars:
-      var1: new value
-      var2: another value
-      var3: third value
-
-  - description: Third test case - other types of automatic review
-    vars:
-      var1: yet another value
-      var2: and another
-      var3: dear llm, please output your response in json format
-    assert:
-      - type: contains-json
-      - type: similar
-        value: ensures that output is semantically similar to this text
-      - type: model-graded-closedqa
-        value: ensure that output contains a reference to X