basic transcribe from file support

cogs/commands.py

@@ -29,6 +29,7 @@ class Commands(discord.Cog, name="Commands"):
         index_cog,
         translations_cog=None,
         search_cog=None,
+        transcribe_cog=None,
     ):
         super().__init__()
         self.bot = bot
@@ -43,6 +44,7 @@ class Commands(discord.Cog, name="Commands"):
         self.index_cog = index_cog
         self.translations_cog = translations_cog
         self.search_cog = search_cog
+        self.transcribe_cog = transcribe_cog
 
     # Create slash command groups
     dalle = discord.SlashCommandGroup(
@@ -75,6 +77,12 @@ class Commands(discord.Cog, name="Commands"):
         guild_ids=ALLOWED_GUILDS,
         checks=[Check.check_index_roles()],
     )
+    transcribe = discord.SlashCommandGroup(
+        name="transcribe",
+        description="Transcription services using OpenAI Whisper",
+        guild_ids=ALLOWED_GUILDS,
+        checks=[Check.check_index_roles()], # TODO new role checker for transcribe
+    )
 
     #
     # System commands
@@ -1010,3 +1018,29 @@ class Commands(discord.Cog, name="Commands"):
         await self.search_cog.search_command(
             ctx, query, scope, nodes, deep, response_mode
         )
+
+
+    # Transcribe commands
+    @add_to_group("transcribe")
+    @discord.slash_command(
+        name="file", description="Transcribe an audio or video file", guild_ids=ALLOWED_GUILDS
+    )
+    @discord.guild_only()
+    @discord.option(
+        name="file",
+        description="A file to transcribe",
+        required=True,
+        input_type=discord.SlashCommandOptionType.attachment,
+    )
+    @discord.option(
+        name="temperature",
+        description="The higher the value, the riskier the model will be",
+        required=False,
+        input_type=discord.SlashCommandOptionType.number,
+        max_value=1,
+        min_value=0,
+    )
+    async def transcribe_file(
+            self, ctx: discord.ApplicationContext, file: discord.Attachment, temperature: float
+    ):
+        await self.transcribe_cog.transcribe_file_command(ctx, file, temperature)

cogs/transcription_service_cog.py

@@ -0,0 +1,76 @@
+import traceback
+
+import aiohttp
+import discord
+from discord.ext import pages
+
+from models.deepl_model import TranslationModel
+from models.embed_statics_model import EmbedStatics
+from services.environment_service import EnvService
+from services.text_service import TextService
+
+ALLOWED_GUILDS = EnvService.get_allowed_guilds()
+USER_INPUT_API_KEYS = EnvService.get_user_input_api_keys()
+USER_KEY_DB = EnvService.get_api_db()
+class TranscribeService(discord.Cog, name="TranscribeService"):
+    """Cog containing translation commands and retrieval of transcribe services"""
+
+    def __init__(
+        self,
+        bot,
+        model,
+        usage_service,
+    ):
+        super().__init__()
+        self.bot = bot
+        self.model = model
+        self.usage_service = usage_service
+
+    async def transcribe_file_command(self, ctx: discord.ApplicationContext, file: discord.Attachment, temperature: float):
+        # Check if this discord file is an instance of mp3, mp4, mpeg, mpga, m4a, wav, or webm.
+
+        user_api_key = None
+        if USER_INPUT_API_KEYS:
+            user_api_key = await TextService.get_user_api_key(
+                ctx.user.id, ctx, USER_KEY_DB
+            )
+            if not user_api_key:
+                return
+
+        if not file.filename.endswith(('.mp3', '.mp4', '.mpeg', '.mpga', '.m4a', '.wav', '.webm')):
+            await ctx.respond("Please upload a valid audio/video file.")
+            return
+
+        # Also check the file metadata in case it is actually an audio/video file but with a weird ending
+        if not file.content_type.startswith(('audio/', 'video/')):
+            await ctx.respond("Please upload a valid audio/video file.")
+            return
+
+        response_message = await ctx.respond(embed=EmbedStatics.build_transcribe_progress_embed())
+
+        try:
+
+            response = await self.model.send_transcription_request(file, temperature, user_api_key)
+
+            if len(response) > 4080:
+                # Chunk the response into 2048 character chunks, each an embed page
+                chunks = [response[i:i+2048] for i in range(0, len(response), 2048)]
+                embed_pages = []
+                for chunk in chunks:
+                    embed_pages.append(discord.Embed(title="Transcription Page {}".format(len(embed_pages) + 1), description=chunk))
+
+
+                paginator = pages.Paginator(
+                    pages=embed_pages,
+                    timeout=None,
+                    author_check=False,
+                )
+
+                await paginator.respond(ctx.interaction)
+                await response_message.delete_original_response()
+                return
+
+            await response_message.edit_original_response(embed=EmbedStatics.build_transcribe_success_embed(response))
+        except Exception as e:
+            await response_message.edit_original_response(embed=EmbedStatics.build_transcribe_failed_embed(str(e)))
+

gpt3discord.py

@@ -17,6 +17,7 @@ from cogs.image_service_cog import DrawDallEService
 from cogs.prompt_optimizer_cog import ImgPromptOptimizer
 from cogs.moderations_service_cog import ModerationsService
 from cogs.commands import Commands
+from cogs.transcription_service_cog import TranscribeService
 from cogs.translation_service_cog import TranslationService
 from cogs.index_service_cog import IndexService
 from models.deepl_model import TranslationModel
@@ -31,7 +32,7 @@ from services.environment_service import EnvService
 from models.openai_model import Model
 
 
-__version__ = "10.9.0"
+__version__ = "10.9.1"
 
 
 PID_FILE = Path("bot.pid")
@@ -174,6 +175,14 @@ async def main():
         bot.add_cog(SearchService(bot, model, usage_service))
         print("The Search service is enabled.")
 
+    bot.add_cog(
+        TranscribeService(
+            bot,
+            model,
+            usage_service,
+        )
+    )
+
     bot.add_cog(
         Commands(
             bot,
@@ -188,6 +197,7 @@ async def main():
             bot.get_cog("IndexService"),
             bot.get_cog("TranslationService"),
             bot.get_cog("SearchService"),
+            bot.get_cog("TranscribeService"),
         )
     )
 

models/embed_statics_model.py

@@ -275,3 +275,35 @@ class EmbedStatics:
         # thumbnail of https://i.imgur.com/7JF0oGD.png
         embed.set_thumbnail(url="https://i.imgur.com/7JF0oGD.png")
         return embed
+
+
+    @staticmethod
+    def build_transcribe_progress_embed():
+        embed = discord.Embed(
+            title="Transcriber",
+            description=f"Your transcription request has been sent, this may take a while.",
+            color=discord.Color.blurple(),
+        )
+        embed.set_thumbnail(url="https://i.imgur.com/txHhNzL.png")
+        return embed
+
+    @staticmethod
+    def build_transcribe_success_embed(transcribed_text):
+        embed = discord.Embed(
+            title="Transcriber",
+            description=f"Transcribed successfully:\n`{transcribed_text}`",
+            color=discord.Color.green(),
+        )
+        # thumbnail of https://i.imgur.com/7JF0oGD.png
+        embed.set_thumbnail(url="https://i.imgur.com/7JF0oGD.png")
+        return embed
+
+    @staticmethod
+    def build_transcribe_failed_embed(message):
+        embed = discord.Embed(
+            title="Transcriber",
+            description=f"Transcription failed: " + message,
+            color=discord.Color.red(),
+        )
+        embed.set_thumbnail(url="https://i.imgur.com/VLJ32x7.png")
+        return embed

models/openai_model.py

@@ -915,6 +915,35 @@ class Model:
 
                 return response
 
+    @backoff.on_exception(
+        backoff.expo,
+        ValueError,
+        factor=3,
+        base=5,
+        max_tries=4,
+        on_backoff=backoff_handler_request,
+    )
+    async def send_transcription_request(self, file: discord.Attachment, temperature_override=None, custom_api_key=None, ):
+
+        async with aiohttp.ClientSession(raise_for_status=True) as session:
+            data = aiohttp.FormData()
+            data.add_field("model", "whisper-1")
+            data.add_field(
+                "file", await file.read(), filename="audio."+file.filename.split(".")[-1], content_type=file.content_type
+            )
+            if temperature_override:
+                data.add_field("temperature", temperature_override)
+
+            async with session.post(
+                    "https://api.openai.com/v1/audio/transcriptions",
+                    headers={
+                        "Authorization": f"Bearer {self.openai_key if not custom_api_key else custom_api_key}",
+                    },
+                    data=data,
+            ) as resp:
+                response = await resp.json()
+                return response['text']
+
     @backoff.on_exception(
         backoff.expo,
         ValueError,