diff --git a/.gitignore b/.gitignore
index 19efb5a940189d778f1f8cbc68122a292e5c5aa7..c5cc22a30ae8b7864be9d01e8f9257ff17febd74 100644
--- a/.gitignore
+++ b/.gitignore
@@ -7,4 +7,5 @@ resource.zip
faq
.gradio
*.old
-get_audio.py
\ No newline at end of file
+get_audio.py
+.vscode
\ No newline at end of file
diff --git a/api.py b/api.py
index a8c67b5dfb54085b7185eec5e9855700db6e88cc..f1fef926229ae330a3f0c52a56367278587a482e 100644
--- a/api.py
+++ b/api.py
@@ -3,10 +3,21 @@ import pickle
from datetime import datetime
from pathlib import Path
-from fish_audio_sdk import Session, TTSRequest
+from fishaudio import FishAudio
+from fishaudio.utils import save
from openai import OpenAI
from tenacity import retry, stop_after_attempt, wait_exponential
+try:
+ from dotenv import load_dotenv
+
+ env_path = Path(os.path.dirname(__file__)) / ".env"
+ if env_path.exists():
+ load_dotenv(env_path)
+except ImportError:
+ pass
+
+# params
INFO_audio_ID_dict = {
"matsu": os.getenv("INFO_audio_matsu"),
"ken": os.getenv("INFO_audio_ken"),
@@ -18,7 +29,7 @@ INFO_audio_ID_dict = {
client = OpenAI(api_key=os.getenv("gpt"))
#
-audio_client = Session(os.getenv("audio"))
+audio_client = FishAudio(api_key=os.getenv("audio"))
PKL_FILE = "./resource/knowledge_data.pkl"
@@ -42,22 +53,11 @@ def save_pkl(file_name, data):
pickle.dump(sorted_data, f)
-try:
- from dotenv import load_dotenv
-
- env_path = Path(os.path.dirname(__file__)) / ".env"
- if env_path.exists():
- load_dotenv(env_path)
-except ImportError:
- pass
-# params
-
-
@retry(
stop=stop_after_attempt(5),
wait=wait_exponential(multiplier=1, min=2, max=10),
)
-def generate_item(prompt, sys_prompt, model="gpt-4.1-mini", temperature=1.0):
+def generate_item(prompt, sys_prompt, model="gpt-5.4-mini", temperature=1.0):
# print("prompt=", prompt)
# print("temperature=", temperature)
response = client.chat.completions.create(
@@ -90,13 +90,20 @@ def get_embedding(text):
def audio_text(text, audio_path, speaker_id):
# TODO change speaker
# print(speaker_id, ":", INFO_audio_ID_dict[speaker_id])
-
- with open(audio_path, "wb") as f:
- for chunk in audio_client.tts(
- TTSRequest(
- reference_id=INFO_audio_ID_dict[speaker_id], text=text, latency="normal"
- ),
- backend="s1",
- ):
- f.write(chunk)
+ audio = audio_client.tts.convert(
+ text=text,
+ reference_id=INFO_audio_ID_dict[speaker_id],
+ format="mp3",
+ model="s2-pro",
+ )
+ save(audio, audio_path)
+
+ # with open(audio_path, "wb") as f:
+ # for chunk in audio_client.tts(
+ # TTSRequest(
+ # reference_id=INFO_audio_ID_dict[speaker_id], text=text, latency="normal"
+ # ),
+ # backend="s1",
+ # ):
+ # f.write(chunk)
return audio_path
diff --git a/app.py b/app.py
index d61c732a33dac9bc0b8d956f9ddfd3a7bbc86ca6..d42bbe7e514ac09e14603fa4815a90a6f5428e6c 100644
--- a/app.py
+++ b/app.py
@@ -79,13 +79,11 @@ def answer_question(
def run_audio_answer():
run_audio_path = audio_text(answer_text, audio_file_path, speaker_flag)
audio_str = base64.b64encode(open(run_audio_path, "rb").read()).decode("utf-8")
- history[-2]["content"] = gr.HTML(
- f"""
+ history[-2]["content"] = gr.HTML(f"""
- """
- )
+ """)
history[-2]["metadata"] = {"title": "🎧 音声返信", "status": "done"}
if not audio_flag:
@@ -202,7 +200,7 @@ with gr.Blocks(title="勉強会", css=custom_css, head=custom_head, js=js_func)
bubble_full_width=False,
avatar_images=(
None,
- "https://cdn.profile-image.st-hatena.com/users/saratoga623/profile.png?1728512391",
+ "https://assets.st-note.com/production/uploads/images/212145872/profile_ec44d8ebf5cfcb95f777ad3dacdeaacf.png?fit=bounds&format=jpeg&quality=85&width=330",
),
)
diff --git a/blog_class.py b/blog_class.py
index b64482e5226605968771dbab0681e3f744c74841..fc5c07fa4f5463f14ac2e1f848d4bc89a0844e18 100644
--- a/blog_class.py
+++ b/blog_class.py
@@ -158,7 +158,7 @@ class knowledge_class:
)
print("user_prompt:", user_prompt)
answer_text = generate_item(
- user_prompt, SYS_Prompt, model="gpt-4.1", temperature=temperature
+ user_prompt, SYS_Prompt, model="gpt-5.4", temperature=temperature
)
audio_filename = save_md_and_get_audio(user_prompt, answer_text)
return answer_text, info_title, audio_filename
@@ -216,7 +216,7 @@ class knowledge_class:
chat_history=chat_history_str, question=new_query
)
- rewrite_question = generate_item(rw_prompt, REWRITE_SYS_Prompt, model="gpt-4.1")
+ rewrite_question = generate_item(rw_prompt, REWRITE_SYS_Prompt, model="gpt-5.4")
print("rewrite_question:", rewrite_question)
# get rewrite question
@@ -244,7 +244,7 @@ class knowledge_class:
answer_text = generate_item(
user_prompt,
SYS_Prompt,
- model="gpt-4.1",
+ model="gpt-5.4",
temperature=temperature,
)
audio_filename = save_md_and_get_audio(user_prompt, answer_text)
diff --git a/design.py b/design.py
index 29932d1d8b14212f07e13164ed4a5bc834c12c1f..422354b7a2ef0c57865c07e103e4a29d48f95af7 100644
--- a/design.py
+++ b/design.py
@@ -70,7 +70,7 @@ function refresh() {
title_html = """
-
+
