User Tools
язык_программирования_python
Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision | ||
|
язык_программирования_python [2026/07/16 09:47] val [ari_rec.py] |
язык_программирования_python [2026/07/17 06:25] (current) val [ari_chat.py] |
||
|---|---|---|---|
| Line 402: | Line 402: | ||
| ==== wav_and_ogg.py ==== | ==== wav_and_ogg.py ==== | ||
| + | |||
| + | * [[Перекодировка звука#Перекодировка в ogg формат]] | ||
| + | |||
| <code> | <code> | ||
| + | $ pip install pydub pip #audioop-lts | ||
| + | </code><code> | ||
| import sys | import sys | ||
| from pydub import AudioSegment | from pydub import AudioSegment | ||
| Line 424: | Line 429: | ||
| # wav_to_ogg(target_file) | # wav_to_ogg(target_file) | ||
| # ogg_to_wav(target_file) | # ogg_to_wav(target_file) | ||
| + | </code><code> | ||
| + | $ python wav_and_ogg.py speech | ||
| </code> | </code> | ||
| ==== speech_and_text.py ==== | ==== speech_and_text.py ==== | ||
| + | |||
| + | * [[Yandex AI]] | ||
| + | |||
| <code> | <code> | ||
| import os | import os | ||
| Line 432: | Line 442: | ||
| import requests | import requests | ||
| - | API_KEY = os.getenv("API_KEY") | + | API_KEY = os.getenv("YANDEX_API_KEY") |
| - | FOLDER_ID = os.getenv("FOLDER_ID") | + | FOLDER_ID = os.getenv("YANDEX_FOLDER_ID") |
| def speech_to_text(file_path): | def speech_to_text(file_path): | ||
| Line 497: | Line 507: | ||
| sdk = AIStudio( | sdk = AIStudio( | ||
| - | folder_id=os.getenv("FOLDER_ID"), | + | folder_id=os.getenv("YANDEX_FOLDER_ID"), |
| - | auth=os.getenv("API_KEY") | + | auth=os.getenv("YANDEX_API_KEY") |
| ) | ) | ||
| messages = [{"role": "system", "text": system_prompt}] | messages = [{"role": "system", "text": system_prompt}] | ||
| Line 696: | Line 706: | ||
| "Ты — ассистент.\n" | "Ты — ассистент.\n" | ||
| "Отвечай кратко.\n" | "Отвечай кратко.\n" | ||
| + | "Не спрашивай когда напомнить.\n" | ||
| ) | ) | ||
| + | COMPRESS_PROMPT = ( | ||
| + | "Ты — инструмент оптимизации контекста. Проанализируй историю диалога. " | ||
| + | "Удали выполненные дела, приветствия и воду. Сформулируй ОДНО лаконичное " | ||
| + | "сообщение от лица пользователя, в котором отражена только актуальная суть " | ||
| + | "и невыполненные задачи/оставшиеся вопросы на текущий момент." | ||
| + | ) | ||
| # Структура: {channel_id: [{"role": "user", "text": "..."}, {"role": "assistant", "text": "..."}]} | # Структура: {channel_id: [{"role": "user", "text": "..."}, {"role": "assistant", "text": "..."}]} | ||
| channel_contexts = {} | channel_contexts = {} | ||
| + | |||
| channel_caller_numbers = {} | channel_caller_numbers = {} | ||
| Line 727: | Line 745: | ||
| def start_recording(channel_id): | def start_recording(channel_id): | ||
| - | print(f"Включаем запись для канала {channel_id} на 5 секунд...") | + | print(f"Включаем запись для канала {channel_id} на 29 секунд...") |
| record_url = f"{BASE_URL}/channels/{channel_id}/record" | record_url = f"{BASE_URL}/channels/{channel_id}/record" | ||
| recording_name = f"rec_{channel_id}" | recording_name = f"rec_{channel_id}" | ||
| Line 734: | Line 752: | ||
| "format": "wav", | "format": "wav", | ||
| "ifExists": "overwrite", | "ifExists": "overwrite", | ||
| - | "maxDuration": 5, | + | "maxDuration": 29, |
| "terminateOn": "#", | "terminateOn": "#", | ||
| "beep": True, | "beep": True, | ||
| Line 777: | Line 795: | ||
| user_request = speech_and_text.speech_to_text(f"{REC_PATH}{recording_name}.ogg") | user_request = speech_and_text.speech_to_text(f"{REC_PATH}{recording_name}.ogg") | ||
| - | print(user_request) | + | print(f"user_request: {user_request}") |
| - | context = get_channel_context(channel_id) | + | if user_request: |
| + | context = get_channel_context(channel_id) | ||
| - | import request_to_agent | + | import request_to_agent |
| - | agent_response = request_to_agent.request_to_agent(user_request, SYSTEM_PROMPT, context) | + | agent_response = request_to_agent.request_to_agent(user_request, SYSTEM_PROMPT, context) |
| - | #agent_response = user_request | + | #agent_response = user_request |
| - | print(agent_response) | + | print(agent_response) |
| - | update_channel_context(channel_id, user_request, agent_response) | + | update_channel_context(channel_id, user_request, agent_response) |
| - | print("Текущий контекст:") | + | print("Текущий контекст:") |
| - | from pprint import pprint | + | from pprint import pprint |
| - | pprint(channel_contexts) | + | pprint(channel_contexts[channel_id]) |
| - | speech_and_text.text_to_speech(agent_response,f"{REC_PATH}{recording_name}.response.ogg") | + | speech_and_text.text_to_speech(agent_response,f"{REC_PATH}{recording_name}.response.ogg") |
| - | wav_and_ogg.ogg_to_wav(f"{REC_PATH}{recording_name}.response") | + | wav_and_ogg.ogg_to_wav(f"{REC_PATH}{recording_name}.response") |
| - | print(f"Запись {recording_name} готова! Проигрываем её обратно в канал {channel_id}...") | + | print(f"Запись {recording_name} готова! Проигрываем её обратно в канал {channel_id}...") |
| - | play_url = f"{BASE_URL}/channels/{channel_id}/play" | + | play_url = f"{BASE_URL}/channels/{channel_id}/play" |
| - | play_payload = { | + | play_payload = { |
| - | #"media": f"recording:{recording_name}" | + | #"media": f"recording:{recording_name}" |
| - | "media": f"recording:{recording_name}.response" | + | "media": f"recording:{recording_name}.response" |
| - | } | + | } |
| - | response = requests.post(play_url, params=play_payload, auth=AUTH) | + | response = requests.post(play_url, params=play_payload, auth=AUTH) |
| - | print("Ответ ARI на воспроизведение:", response.status_code) | + | print("Ответ ARI на воспроизведение:", response.status_code) |
| # Сценарий 3: Воспроизведение завершено | # Сценарий 3: Воспроизведение завершено | ||
| Line 821: | Line 840: | ||
| caller_number = channel_caller_numbers[channel_id] | caller_number = channel_caller_numbers[channel_id] | ||
| print(f"Номер звонящего в StasisEnd: номер {caller_number} канал {channel_id}") | print(f"Номер звонящего в StasisEnd: номер {caller_number} канал {channel_id}") | ||
| + | |||
| + | print(f"Сжатие контекста") | ||
| + | |||
| + | import request_to_agent | ||
| + | agent_response = request_to_agent.request_to_agent("Сожми контекст", COMPRESS_PROMPT, context) | ||
| + | print(agent_response) | ||
| + | context = [{"role": "user", "text": agent_response}] | ||
| print(f"Сохранение контекста и удаление медиа файлов") | print(f"Сохранение контекста и удаление медиа файлов") | ||
язык_программирования_python.1784184466.txt.gz · Last modified: 2026/07/16 09:47 by val
Page Tools
Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
