lyrics-transcriber 0.18.0__tar.gz → 0.19.2__tar.gz

{lyrics_transcriber-0.18.0 → lyrics_transcriber-0.19.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: lyrics-transcriber
-Version: 0.18.0
+Version: 0.19.2
 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

{lyrics_transcriber-0.18.0 → lyrics_transcriber-0.19.2}/lyrics_transcriber/audioshake_transcriber.py

@@ -5,13 +5,15 @@ import json
 
 
 class AudioShakeTranscriber:
-    def __init__(self, api_token, logger):
+    def __init__(self, api_token, logger, output_prefix):
         self.api_token = api_token
         self.base_url = "https://groovy.audioshake.ai"
         self.logger = logger
+        self.output_prefix = output_prefix
 
-    def transcribe(self, audio_filepath):
-        self.logger.info(f"Transcribing {audio_filepath} using AudioShake API")
+    def start_transcription(self, audio_filepath):
+        """Starts the transcription job and returns the job ID without waiting for completion"""
+        self.logger.info(f"Starting transcription for {audio_filepath} using AudioShake API")
 
         # Step 1: Upload the audio file
         asset_id = self._upload_file(audio_filepath)
@@ -21,6 +23,12 @@ class AudioShakeTranscriber:
         job_id = self._create_job(asset_id)
         self.logger.info(f"Job created successfully. Job ID: {job_id}")
 
+        return job_id
+
+    def get_transcription_result(self, job_id):
+        """Gets the results for a previously started job"""
+        self.logger.info(f"Getting results for job ID: {job_id}")
+
         # Step 3: Wait for the job to complete and get the results
         result = self._get_job_result(job_id)
         self.logger.info(f"Job completed. Processing results...")
@@ -28,6 +36,11 @@ class AudioShakeTranscriber:
         # Step 4: Process the result and return in the required format
         return self._process_result(result)
 
+    def transcribe(self, audio_filepath):
+        """Original method now just combines the two steps"""
+        job_id = self.start_transcription(audio_filepath)
+        return self.get_transcription_result(job_id)
+
     def _upload_file(self, filepath):
         self.logger.info(f"Uploading {filepath} to AudioShake")
         url = f"{self.base_url}/upload"
@@ -76,13 +89,10 @@ class AudioShakeTranscriber:
         output_assets = job_data.get("outputAssets", [])
         self.logger.debug(f"Output assets: {output_assets}")
 
-        output_asset = next((asset for asset in output_assets if asset["name"] == "transcription.json"), None)
-        if not output_asset:
-            self.logger.warning("'transcription.json' not found, looking for 'alignment.json'")
-            output_asset = next((asset for asset in output_assets if asset["name"] == "alignment.json"), None)
+        output_asset = next((asset for asset in output_assets if asset["name"] == "alignment.json"), None)
 
         if not output_asset:
-            self.logger.error("Neither 'transcription.json' nor 'alignment.json' found in job results")
+            self.logger.error("'alignment.json' found in job results")
             self.logger.error(f"Available output assets: {[asset['name'] for asset in output_assets]}")
             raise Exception("Required output not found in job results")
 
@@ -103,4 +113,10 @@ class AudioShakeTranscriber:
             if "text" not in segment:
                 segment["text"] = " ".join(word["text"] for word in segment["words"])
 
+        transcription_data["output_filename"] = self.get_output_filename(" (AudioShake)")
+
         return transcription_data
+
+    def get_output_filename(self, suffix):
+        """Generate consistent filename with (Purpose) suffix pattern"""
+        return f"{self.output_prefix}{suffix}"

{lyrics_transcriber-0.18.0 → lyrics_transcriber-0.19.2}/lyrics_transcriber/transcriber.py

@@ -66,7 +66,7 @@ class LyricsTranscriber:
         self.openai_api_key = os.getenv("OPENAI_API_KEY", default=openai_api_key)
         self.genius_api_token = os.getenv("GENIUS_API_TOKEN", default=genius_api_token)
         self.spotify_cookie = os.getenv("SPOTIFY_COOKIE_SP_DC", default=spotify_cookie)
-        self.audioshake_api_token = os.getenv("AUDIOSHAKE_TOKEN", default=audioshake_api_token)
+        self.audioshake_api_token = os.getenv("AUDIOSHAKE_API_TOKEN", default=audioshake_api_token)
 
         self.transcription_model = transcription_model
         self.llm_model = llm_model
@@ -102,7 +102,7 @@ class LyricsTranscriber:
 
             self.openai_client.log = self.log_level
         else:
-            self.logger.error("No OpenAI API key found, no correction will be applied to transcription")
+            self.logger.warning("No OpenAI API key found, no correction will be applied to transcription")
 
         self.render_video = render_video
         self.video_resolution = video_resolution
@@ -137,10 +137,18 @@ class LyricsTranscriber:
                 raise FileNotFoundError(f"video_background is not a valid file path: {self.video_background_image}")
 
         self.outputs = {
-            "transcription_data_dict": None,
-            "transcription_data_filepath": None,
-            "transcribed_lyrics_text": None,
-            "transcribed_lyrics_text_filepath": None,
+            "transcription_data_dict_whisper": None,
+            "transcription_data_whisper_filepath": None,
+            "transcribed_lyrics_text_whisper": None,
+            "transcribed_lyrics_text_whisper_filepath": None,
+            "transcription_data_dict_audioshake": None,
+            "transcription_data_audioshake_filepath": None,
+            "transcribed_lyrics_text_audioshake": None,
+            "transcribed_lyrics_text_audioshake_filepath": None,
+            "transcription_data_dict_primary": None,
+            "transcription_data_primary_filepath": None,
+            "transcribed_lyrics_text_primary": None,
+            "transcribed_lyrics_text_primary_filepath": None,
             "genius_lyrics_text": None,
             "genius_lyrics_filepath": None,
             "spotify_lyrics_data_dict": None,
@@ -166,10 +174,15 @@ class LyricsTranscriber:
 
         self.create_folders()
 
+        self.output_prefix = f"{artist} - {title}"
+
     def generate(self):
+        self.logger.debug(f"Starting generate() with cache_dir: {self.cache_dir} and output_dir: {self.output_dir}")
+
         self.logger.debug(f"audio_filepath is set: {self.audio_filepath}, beginning initial whisper transcription")
 
         self.transcribe()
+
         self.write_transcribed_lyrics_plain_text()
 
         self.write_genius_lyrics_file()
@@ -183,7 +196,7 @@ class LyricsTranscriber:
             self.write_corrected_lyrics_plain_text()
         else:
             self.logger.warning("Skipping LLM correction as no OpenAI client is available")
-            self.outputs["corrected_lyrics_data_dict"] = self.outputs["transcription_data_dict"]
+            self.outputs["corrected_lyrics_data_dict"] = self.outputs["transcription_data_dict_primary"]
             self.write_corrected_lyrics_plain_text()
 
         self.calculate_singing_percentage()
@@ -208,11 +221,15 @@ class LyricsTranscriber:
             self.output_dir = os.getcwd()
 
         self.logger.debug(f"copying temporary files to output dir: {self.output_dir}")
-
-        for key in self.outputs:
+        self.logger.debug("Files to copy:")
+        for key, value in self.outputs.items():
             if key.endswith("_filepath"):
-                if self.outputs[key] and os.path.isfile(self.outputs[key]):
-                    shutil.copy(self.outputs[key], self.output_dir)
+                self.logger.debug(f"  {key}: {value}")
+                if value and os.path.isfile(value):
+                    self.logger.debug(f"    File exists, copying to {self.output_dir}")
+                    shutil.copy(value, self.output_dir)
+                else:
+                    self.logger.debug(f"    File doesn't exist or is None")
 
         self.outputs["output_dir"] = self.output_dir
 
@@ -232,9 +249,7 @@ class LyricsTranscriber:
                 continue
 
             if self.openai_client:
-                data_input_str = (
-                    f'Data input 1:\n{self.outputs["transcribed_lyrics_text"]}\nData input 2:\n{self.outputs[online_lyrics_text_key]}\n'
-                )
+                data_input_str = f'Data input 1:\n{self.outputs["transcribed_lyrics_text_primary"]}\nData input 2:\n{self.outputs[online_lyrics_text_key]}\n'
 
                 self.logger.debug(f"making API call to LLM model {self.llm_model} to validate {online_lyrics_source} lyrics match")
                 response = self.openai_client.chat.completions.create(
@@ -263,7 +278,7 @@ class LyricsTranscriber:
             else:
                 # Fallback primitive word matching
                 self.logger.debug(f"Using primitive word matching to validate {online_lyrics_source} lyrics match")
-                transcribed_words = set(self.outputs["transcribed_lyrics_text"].split())
+                transcribed_words = set(self.outputs["transcribed_lyrics_text_primary"].split())
                 online_lyrics_words = set(self.outputs[online_lyrics_text_key].split())
                 common_words = transcribed_words & online_lyrics_words
                 match_percentage = len(common_words) / len(online_lyrics_words) * 100
@@ -294,7 +309,7 @@ class LyricsTranscriber:
 
         self.logger.debug("write_corrected_lyrics_data_file initiating OpenAI client")
 
-        corrected_lyrics_data_json_cache_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + "-corrected.json")
+        corrected_lyrics_data_json_cache_filepath = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics Corrected).json"))
 
         if os.path.isfile(corrected_lyrics_data_json_cache_filepath):
             self.logger.debug(
@@ -312,7 +327,7 @@ class LyricsTranscriber:
 
         if not reference_lyrics:
             self.logger.warning("No reference lyrics found from Genius or Spotify. Skipping LLM correction.")
-            self.outputs["corrected_lyrics_data_dict"] = self.outputs["transcription_data_dict"]
+            self.outputs["corrected_lyrics_data_dict"] = self.outputs["transcription_data_dict_primary"]
             return
 
         self.logger.debug(
@@ -331,11 +346,9 @@ class LyricsTranscriber:
         # TODO: Possibly add a step after segment-based correct to get the LLM to self-analyse the diff
 
         self.outputs["llm_transcript"] = ""
-        self.outputs["llm_transcript_filepath"] = os.path.join(
-            self.cache_dir, "lyrics-" + self.get_song_slug() + "-llm-correction-transcript.txt"
-        )
+        self.outputs["llm_transcript_filepath"] = os.path.join(self.cache_dir, self.get_output_filename(" (LLM Transcript).txt"))
 
-        total_segments = len(self.outputs["transcription_data_dict"]["segments"])
+        total_segments = len(self.outputs["transcription_data_dict_primary"]["segments"])
         self.logger.info(f"Beginning correction using LLM, total segments: {total_segments}")
 
         with open(self.outputs["llm_transcript_filepath"], "a", buffering=1, encoding="utf-8") as llm_transcript_file:
@@ -345,7 +358,7 @@ class LyricsTranscriber:
             self.outputs["llm_transcript"] += llm_transcript_header
             llm_transcript_file.write(llm_transcript_header)
 
-            for segment in self.outputs["transcription_data_dict"]["segments"]:
+            for segment in self.outputs["transcription_data_dict_primary"]["segments"]:
                 # # Don't waste OpenAI dollars when testing!
                 # if segment["id"] > 10:
                 #     continue
@@ -371,7 +384,7 @@ class LyricsTranscriber:
                     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"]:
+                for next_segment in self.outputs["transcription_data_dict_primary"]["segments"]:
                     if next_segment["id"] in (segment["id"] + 1, segment["id"] + 2):
                         upcoming_two_uncorrected_lines += next_segment["text"].strip() + "\n"
 
@@ -466,7 +479,9 @@ class LyricsTranscriber:
         if self.outputs["corrected_lyrics_data_dict"]:
             self.logger.debug(f"corrected_lyrics_data_dict exists, writing plain text lyrics file")
 
-            corrected_lyrics_text_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + "-corrected.txt")
+            corrected_lyrics_text_filepath = os.path.join(
+                self.cache_dir, self.get_output_filename(" (Lyrics Corrected).txt")  # Updated to use consistent naming
+            )
             self.outputs["corrected_lyrics_text_filepath"] = corrected_lyrics_text_filepath
 
             self.outputs["corrected_lyrics_text"] = ""
@@ -475,7 +490,7 @@ class LyricsTranscriber:
             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")
+                    f.write(corrected_segment["text"].strip() + "\n")
 
     def write_spotify_lyrics_data_file(self):
         if self.spotify_cookie and self.song_known:
@@ -484,7 +499,9 @@ class LyricsTranscriber:
             self.logger.warning(f"skipping spotify fetch as not all spotify params were set")
             return
 
-        spotify_lyrics_data_json_cache_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + "-spotify.json")
+        spotify_lyrics_data_json_cache_filepath = os.path.join(
+            self.cache_dir, self.get_output_filename(" (Lyrics Spotify).json")  # Updated to use consistent naming
+        )
 
         if os.path.isfile(spotify_lyrics_data_json_cache_filepath):
             self.logger.debug(
@@ -531,7 +548,9 @@ class LyricsTranscriber:
         if self.outputs["spotify_lyrics_data_dict"]:
             self.logger.debug(f"spotify_lyrics data found, checking/writing plain text lyrics file")
 
-            spotify_lyrics_text_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + "-spotify.txt")
+            spotify_lyrics_text_filepath = os.path.join(
+                self.cache_dir, self.get_output_filename(" (Lyrics Spotify).txt")  # Updated to use consistent naming
+            )
             self.outputs["spotify_lyrics_text_filepath"] = spotify_lyrics_text_filepath
 
             lines = self.outputs["spotify_lyrics_data_dict"]["lyrics"]["lines"]
@@ -561,8 +580,9 @@ class LyricsTranscriber:
             self.logger.warning(f"skipping genius fetch as not all genius params were set")
             return
 
-        genius_lyrics_cache_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + "-genius.txt")
+        genius_lyrics_cache_filepath = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics Genius).txt"))
 
+        # Check cache first
         if os.path.isfile(genius_lyrics_cache_filepath):
             self.logger.debug(f"found existing file at genius_lyrics_cache_filepath, reading: {genius_lyrics_cache_filepath}")
 
@@ -570,15 +590,21 @@ class LyricsTranscriber:
                 self.outputs["genius_lyrics_filepath"] = genius_lyrics_cache_filepath
                 self.outputs["genius_lyrics_text"] = cached_lyrics.read()
                 return
-
         self.logger.debug(f"no cached lyrics found at genius_lyrics_cache_filepath: {genius_lyrics_cache_filepath}, fetching from Genius")
-        genius = lyricsgenius.Genius(self.genius_api_token, verbose=(self.log_level == logging.DEBUG))
+
+        # Initialize Genius with better defaults
+        genius = lyricsgenius.Genius(
+            self.genius_api_token,
+            verbose=(self.log_level == logging.DEBUG),
+            remove_section_headers=True,
+        )
 
         try:
             song = self.fetch_genius_lyrics(genius, self.title, self.artist)
             if song is None:
                 self.logger.warning(f'Could not find lyrics on Genius for "{self.title}" by {self.artist}')
-                return
+                return None
+
             lyrics = self.clean_genius_lyrics(song.lyrics)
 
             self.logger.debug(f"writing clean lyrics to genius_lyrics_cache_filepath: {genius_lyrics_cache_filepath}")
@@ -587,6 +613,8 @@ class LyricsTranscriber:
 
             self.outputs["genius_lyrics_filepath"] = genius_lyrics_cache_filepath
             self.outputs["genius_lyrics_text"] = lyrics
+            return lyrics.split("\n")  # Return lines like write_lyrics_from_genius
+
         except requests.exceptions.RequestException as e:
             self.logger.error(f"Failed to fetch lyrics from Genius after multiple retries: {e}")
             raise
@@ -594,8 +622,13 @@ class LyricsTranscriber:
     def clean_genius_lyrics(self, lyrics):
         lyrics = lyrics.replace("\\n", "\n")
         lyrics = re.sub(r"You might also like", "", lyrics)
-        # Remove the song name and word "Lyrics" if this has a non-newline char at the start
-        lyrics = re.sub(r".*?Lyrics([A-Z])", r"\1", lyrics)
+        lyrics = re.sub(
+            r".*?Lyrics([A-Z])", r"\1", lyrics
+        )  # Remove the song name and word "Lyrics" if this has a non-newline char at the start
+        lyrics = re.sub(r"^[0-9]* Contributors.*Lyrics", "", lyrics)  # Remove this example: 27 ContributorsSex Bomb Lyrics
+        lyrics = re.sub(
+            r"See.*Live.*Get tickets as low as \$[0-9]+", "", lyrics
+        )  # Remove this example: See Tom Jones LiveGet tickets as low as $71
         lyrics = re.sub(r"[0-9]+Embed$", "", lyrics)  # Remove the word "Embed" at end of line with preceding numbers if found
         lyrics = re.sub(r"(\S)Embed$", r"\1", lyrics)  # Remove the word "Embed" if it has been tacked onto a word at the end of a line
         lyrics = re.sub(r"^Embed$", r"", lyrics)  # Remove the word "Embed" if it has been tacked onto a word at the end of a line
@@ -605,7 +638,9 @@ class LyricsTranscriber:
 
     def calculate_singing_percentage(self):
         # Calculate total seconds of singing using timings from whisper transcription results
-        total_singing_duration = sum(segment["end"] - segment["start"] for segment in self.outputs["transcription_data_dict"]["segments"])
+        total_singing_duration = sum(
+            segment["end"] - segment["start"] for segment in self.outputs["transcription_data_dict_primary"]["segments"]
+        )
 
         self.logger.debug(f"calculated total_singing_duration: {int(total_singing_duration)} seconds, now running ffprobe")
 
@@ -635,7 +670,7 @@ class LyricsTranscriber:
     # then loops over each word and writes all words with MidiCo segment start/end formatting
     # and word-level timestamps to a MidiCo-compatible LRC file
     def write_midico_lrc_file(self):
-        self.outputs["midico_lrc_filepath"] = self.get_cache_filepath(".lrc")
+        self.outputs["midico_lrc_filepath"] = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics Corrected).lrc"))
 
         lrc_filename = self.outputs["midico_lrc_filepath"]
         self.logger.debug(f"writing midico formatted word timestamps to LRC file: {lrc_filename}")
@@ -652,7 +687,7 @@ class LyricsTranscriber:
                     f.write(line)
 
     def create_screens(self):
-        self.logger.debug("create_screens beginning generation of screens from whisper results")
+        self.logger.debug("create_screens beginning generation of screens from transcription results")
         screens: List[subtitles.LyricsScreen] = []
         screen: Optional[subtitles.LyricsScreen] = None
 
@@ -692,9 +727,15 @@ class LyricsTranscriber:
                     self.logger.debug("Reset current line")
 
                 current_line_text += (" " if current_line_text else "") + word["text"]
+
+                # fmt: off
                 lyric_segment = subtitles.LyricSegment(
-                    text=word["text"], ts=timedelta(seconds=word["start"]), end_ts=timedelta(seconds=word["end"])
+                    text=word["text"], 
+                    ts=timedelta(seconds=word["start"]), 
+                    end_ts=timedelta(seconds=word["end"])
                 )
+                # fmt: on
+
                 current_line.segments.append(lyric_segment)
                 self.logger.debug(f"Added word to current line. Current line: '{current_line_text}'")
 
@@ -706,13 +747,13 @@ class LyricsTranscriber:
         return screens
 
     def write_ass_file(self):
-        self.outputs["ass_subtitles_filepath"] = self.get_cache_filepath(".ass")
+        self.outputs["ass_subtitles_filepath"] = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics Corrected).ass"))
 
         ass_filepath = self.outputs["ass_subtitles_filepath"]
         self.logger.debug(f"writing ASS formatted subtitle file: {ass_filepath}")
 
-        intial_screens = self.create_screens()
-        screens = subtitles.set_segment_end_times(intial_screens, int(self.outputs["song_duration"]))
+        initial_screens = self.create_screens()
+        screens = subtitles.set_segment_end_times(initial_screens, int(self.outputs["song_duration"]))
         screens = subtitles.set_screen_start_times(screens)
         lyric_subtitles_ass = subtitles.create_styled_subtitles(screens, self.video_resolution_num, self.font_size)
         lyric_subtitles_ass.write(ass_filepath)
@@ -831,22 +872,29 @@ class LyricsTranscriber:
         return formatted_time
 
     def write_transcribed_lyrics_plain_text(self):
-        if self.outputs["transcription_data_dict"]:
-            transcription_cache_suffix = "-audioshake-transcribed.txt" if self.audioshake_api_token else "-whisper-transcribed.txt"
-            self.logger.debug(f"transcription_cache_suffix: {transcription_cache_suffix}")
-
-            transcribed_lyrics_text_filepath = os.path.join(self.cache_dir, "lyrics-" + self.get_song_slug() + transcription_cache_suffix)
-            self.outputs["transcribed_lyrics_text_filepath"] = transcribed_lyrics_text_filepath
+        if self.outputs["transcription_data_dict_whisper"]:
+            transcribed_lyrics_text_whisper_filepath = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics Whisper).txt"))
+            self.logger.debug(f"Setting Whisper text filepath to: {transcribed_lyrics_text_whisper_filepath}")
+            self.outputs["transcribed_lyrics_text_whisper_filepath"] = transcribed_lyrics_text_whisper_filepath
+            self.outputs["transcribed_lyrics_text_whisper"] = ""
+
+            self.logger.debug(f"Writing Whisper lyrics to: {transcribed_lyrics_text_whisper_filepath}")
+            with open(transcribed_lyrics_text_whisper_filepath, "w", encoding="utf-8") as f:
+                for segment in self.outputs["transcription_data_dict_whisper"]["segments"]:
+                    self.outputs["transcribed_lyrics_text_whisper"] += segment["text"] + "\n"
+                    f.write(segment["text"].strip() + "\n")
+            self.logger.debug(f"Finished writing Whisper lyrics, file exists: {os.path.exists(transcribed_lyrics_text_whisper_filepath)}")
 
-            self.outputs["transcribed_lyrics_text"] = ""
+        if self.outputs["transcription_data_dict_audioshake"]:
+            transcribed_lyrics_text_audioshake_filepath = os.path.join(self.cache_dir, self.get_output_filename(" (Lyrics AudioShake).txt"))
+            self.outputs["transcribed_lyrics_text_audioshake_filepath"] = transcribed_lyrics_text_audioshake_filepath
+            self.outputs["transcribed_lyrics_text_audioshake"] = ""
 
-            self.logger.debug(f"writing lyrics plain text to transcribed_lyrics_text_filepath: {transcribed_lyrics_text_filepath}")
-            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"
+            self.logger.debug(f"Writing AudioShake lyrics to: {transcribed_lyrics_text_audioshake_filepath}")
+            with open(transcribed_lyrics_text_audioshake_filepath, "w", encoding="utf-8") as f:
+                for segment in self.outputs["transcription_data_dict_audioshake"]["segments"]:
+                    self.outputs["transcribed_lyrics_text_audioshake"] += segment["text"] + "\n"
                     f.write(segment["text"].strip() + "\n")
-        else:
-            raise Exception("Cannot write transcribed lyrics plain text as transcription_data_dict is not set")
 
     def find_best_split_point(self, text, max_length):
         self.logger.debug(f"Finding best split point for text: '{text}' (max_length: {max_length})")
@@ -949,51 +997,111 @@ class LyricsTranscriber:
         return new_segments
 
     def transcribe(self):
-        transcription_cache_suffix = "-audioshake" if self.audioshake_api_token else "-whisper"
-        self.outputs["transcription_data_filepath"] = self.get_cache_filepath(f"{transcription_cache_suffix}.json")
-
-        transcription_cache_filepath = self.outputs["transcription_data_filepath"]
-        if os.path.isfile(transcription_cache_filepath):
-            self.logger.debug(f"transcribe found existing file at transcription_cache_filepath, reading: {transcription_cache_filepath}")
-            with open(transcription_cache_filepath, "r") as cache_file:
-                self.outputs["transcription_data_dict"] = json.load(cache_file)
-                return
+        # Check cache first
+        transcription_cache_filepath_whisper = self.get_cache_filepath(" (Lyrics Whisper).json")
+        transcription_cache_filepath_audioshake = self.get_cache_filepath(" (Lyrics AudioShake).json")
+
+        self.logger.debug(f"Cache directory: {self.cache_dir}")
+        self.logger.debug(f"Output directory: {self.output_dir}")
+
+        if os.path.isfile(transcription_cache_filepath_whisper):
+            self.logger.debug(f"Found existing Whisper transcription, reading: {transcription_cache_filepath_whisper}")
+            with open(transcription_cache_filepath_whisper, "r") as cache_file:
+                self.outputs["transcription_data_dict_whisper"] = json.load(cache_file)
+                self.outputs["transcription_data_whisper_filepath"] = transcription_cache_filepath_whisper
+                self.logger.debug(f"Loaded Whisper data and set filepath to: {self.outputs['transcription_data_whisper_filepath']}")
+
+        if os.path.isfile(transcription_cache_filepath_audioshake):
+            self.logger.debug(f"Found existing AudioShake transcription, reading: {transcription_cache_filepath_audioshake}")
+            with open(transcription_cache_filepath_audioshake, "r") as cache_file:
+                self.outputs["transcription_data_dict_audioshake"] = json.load(cache_file)
+                self.outputs["transcription_data_audioshake_filepath"] = transcription_cache_filepath_audioshake
+
+        # If we have both cached transcriptions, set primary and return early
+        if self.outputs["transcription_data_dict_whisper"] and self.outputs["transcription_data_dict_audioshake"]:
+            self.set_primary_transcription()
+            return
+        # If we have Whisper cached and AudioShake isn't available, set primary and return early
+        elif self.outputs["transcription_data_dict_whisper"] and not self.audioshake_api_token:
+            self.set_primary_transcription()
+            return
 
-        if self.audioshake_api_token:
-            self.logger.debug(f"Using AudioShake API for transcription")
+        # Continue with transcription for any missing data...
+        audioshake_job_id = None
+        if self.audioshake_api_token and not self.outputs["transcription_data_dict_audioshake"]:
+            self.logger.debug(f"Starting AudioShake transcription")
             from .audioshake_transcriber import AudioShakeTranscriber
 
-            audioshake = AudioShakeTranscriber(self.audioshake_api_token, logger=self.logger)
-            transcription_data = audioshake.transcribe(self.audio_filepath)
-        else:
+            audioshake = AudioShakeTranscriber(api_token=self.audioshake_api_token, logger=self.logger, output_prefix=self.output_prefix)
+            audioshake_job_id = audioshake.start_transcription(self.audio_filepath)
+
+        # Run Whisper transcription if needed while AudioShake processes
+        if not self.outputs["transcription_data_dict_whisper"]:
             self.logger.debug(f"Using Whisper for transcription with model: {self.transcription_model}")
             audio = whisper.load_audio(self.audio_filepath)
             model = whisper.load_model(self.transcription_model, device="cpu")
-            transcription_data = whisper.transcribe(model, audio, language="en", beam_size=5, temperature=0.2, best_of=5)
-            
-            # auditok is needed for voice activity detection, but it has OS package dependencies that are hard to install on some platforms
-            # transcription_data = whisper.transcribe(model, audio, language="en", vad="auditok", beam_size=5, temperature=0.2, best_of=5)
+            whisper_data = whisper.transcribe(model, audio, language="en", beam_size=5, temperature=0.2, best_of=5)
 
             # Remove segments with no words, only music
-            transcription_data["segments"] = [segment for segment in transcription_data["segments"] if segment["text"].strip() != "Music"]
-            self.logger.debug(f"Removed 'Music' segments. Remaining segments: {len(transcription_data['segments'])}")
+            whisper_data["segments"] = [segment for segment in whisper_data["segments"] if segment["text"].strip() != "Music"]
+            self.logger.debug(f"Removed 'Music' segments. Remaining segments: {len(whisper_data['segments'])}")
 
             # Split long segments
             self.logger.debug("Starting to split long segments")
-            transcription_data["segments"] = self.split_long_segments(transcription_data["segments"], max_length=36)
-            self.logger.debug(f"Finished splitting segments. Total segments after splitting: {len(transcription_data['segments'])}")
-
-        self.logger.debug(f"writing transcription data JSON to cache file: {transcription_cache_filepath}")
-        with open(transcription_cache_filepath, "w") as cache_file:
-            json.dump(transcription_data, cache_file, indent=4)
+            whisper_data["segments"] = self.split_long_segments(whisper_data["segments"], max_length=36)
+            self.logger.debug(f"Finished splitting segments. Total segments after splitting: {len(whisper_data['segments'])}")
+
+            # Store Whisper results
+            self.outputs["transcription_data_dict_whisper"] = whisper_data
+            self.outputs["transcription_data_whisper_filepath"] = transcription_cache_filepath_whisper
+            with open(transcription_cache_filepath_whisper, "w") as cache_file:
+                json.dump(whisper_data, cache_file, indent=4)
+
+        # Now that Whisper is done, get AudioShake results if available
+        if audioshake_job_id:
+            self.logger.debug("Getting AudioShake results")
+            audioshake_data = audioshake.get_transcription_result(audioshake_job_id)
+            self.outputs["transcription_data_dict_audioshake"] = audioshake_data
+            self.outputs["transcription_data_audioshake_filepath"] = transcription_cache_filepath_audioshake
+            with open(transcription_cache_filepath_audioshake, "w") as cache_file:
+                json.dump(audioshake_data, cache_file, indent=4)
+
+        # Set the primary transcription source
+        self.set_primary_transcription()
+
+        # Write the text files
+        self.write_transcribed_lyrics_plain_text()
 
-        self.outputs["transcription_data_dict"] = transcription_data
+    def set_primary_transcription(self):
+        """Set the primary transcription source (AudioShake if available, otherwise Whisper)"""
+        if self.outputs["transcription_data_dict_audioshake"]:
+            self.logger.info("Using AudioShake as primary transcription source")
+            self.outputs["transcription_data_dict_primary"] = self.outputs["transcription_data_dict_audioshake"]
+            self.outputs["transcription_data_primary_filepath"] = self.outputs["transcription_data_audioshake_filepath"]
+
+            # Set the primary text content
+            if "transcribed_lyrics_text_audioshake" not in self.outputs or not self.outputs["transcribed_lyrics_text_audioshake"]:
+                self.outputs["transcribed_lyrics_text_audioshake"] = "\n".join(
+                    segment["text"].strip() for segment in self.outputs["transcription_data_dict_audioshake"]["segments"]
+                )
+            self.outputs["transcribed_lyrics_text_primary"] = self.outputs["transcribed_lyrics_text_audioshake"]
+            self.outputs["transcribed_lyrics_text_primary_filepath"] = self.outputs["transcribed_lyrics_text_audioshake_filepath"]
+        else:
+            self.logger.info("Using Whisper as primary transcription source")
+            self.outputs["transcription_data_dict_primary"] = self.outputs["transcription_data_dict_whisper"]
+            self.outputs["transcription_data_primary_filepath"] = self.outputs["transcription_data_whisper_filepath"]
+
+            # Set the primary text content
+            if "transcribed_lyrics_text_whisper" not in self.outputs or not self.outputs["transcribed_lyrics_text_whisper"]:
+                self.outputs["transcribed_lyrics_text_whisper"] = "\n".join(
+                    segment["text"].strip() for segment in self.outputs["transcription_data_dict_whisper"]["segments"]
+                )
+            self.outputs["transcribed_lyrics_text_primary"] = self.outputs["transcribed_lyrics_text_whisper"]
+            self.outputs["transcribed_lyrics_text_primary_filepath"] = self.outputs["transcribed_lyrics_text_whisper_filepath"]
 
     def get_cache_filepath(self, extension):
-        filename = os.path.split(self.audio_filepath)[1]
-        filename_slug = slugify.slugify(filename, lowercase=False)
-        hash_value = self.get_file_hash(self.audio_filepath)
-        cache_filepath = os.path.join(self.cache_dir, filename_slug + "_" + hash_value + extension)
+        # Instead of using slugify and hash, use the consistent naming pattern
+        cache_filepath = os.path.join(self.cache_dir, self.get_output_filename(extension))
         self.logger.debug(f"get_cache_filepath returning cache_filepath: {cache_filepath}")
         return cache_filepath
 
@@ -1014,3 +1122,7 @@ class LyricsTranscriber:
 
         if self.output_dir is not None:
             os.makedirs(self.output_dir, exist_ok=True)
+
+    def get_output_filename(self, suffix):
+        """Generate consistent filename with (Purpose) suffix pattern"""
+        return f"{self.output_prefix}{suffix}"

{lyrics_transcriber-0.18.0 → lyrics_transcriber-0.19.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "lyrics-transcriber"
-version = "0.18.0"
+version = "0.19.2"
 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"