60db TTS provider integration

.env.template

@@ -17,9 +17,23 @@ AZURE_OPENAI_MODEL=
 AZUREAI_API_KEY=
 AZUREAI_REGION=
 ################################################################################
-### ELEVENLABS API
+### TTS PROVIDER
+## Which text-to-speech provider to use: elevenlabs | sixtydb
+## Keys below are only required for the selected provider.
+################################################################################
+TTS_PROVIDER=elevenlabs
+################################################################################
+### ELEVENLABS API (used when TTS_PROVIDER=elevenlabs)
 ## Eleven Labs Default Voice IDs
 ## https://elevenlabs.io/docs/voicelab/pre-made-voices
 ################################################################################
 ELEVENLABS_API_KEY=
-ELEVENLABS_VOICE_ID=
+ELEVENLABS_VOICE_ID=
+################################################################################
+### 60DB API (used when TTS_PROVIDER=sixtydb)
+## Dashboard: https://app.60db.ai  |  Docs: https://docs.60db.ai
+## Voices: GET https://api.60db.ai/voices (Bearer auth)
+## 60db default voice: fbb75ed2-975a-40c7-9e06-38e30524a9a1
+################################################################################
+SIXTYDB_API_KEY=
+SIXTYDB_VOICE_ID=

README.md

@@ -1,14 +1,14 @@
 # AI Assistant with fast Speech-to-Speech
 
-This project is a Text or Voice Input AI Assistant that uses the OpenAI API to generate responses and the ElevenLabs Text-to-Speech API to convert the responses into audio quickly.
+This project is a Text or Voice Input AI Assistant that uses the OpenAI API to generate responses and the ElevenLabs or 60db Text-to-Speech APIs to convert the responses into audio quickly.
 
 ## Features
 
 - Real-time speech-to-speech conversation
 - Customizable voice selection
+- Switchable TTS provider: ElevenLabs or 60db (`TTS_PROVIDER` in `.env`)
 - Easy-to-use command-line interface
-- Powered by OpenAI and ElevenLabs
-- **Paid ElevenLabs subscription required**
+- Powered by OpenAI, ElevenLabs and 60db
 
 # Getting Started
 
@@ -18,7 +18,9 @@ This project is a Text or Voice Input AI Assistant that uses the OpenAI API to g
 
 - Get your [OpenAI API Key](https://platform.openai.com/api-keys)
 
-- Get your [ElevenLabs API Key](https://elevenlabs.io/app/subscription)
+- Get your [ElevenLabs API Key](https://elevenlabs.io/app/subscription) **or** your [60db API Key](https://app.60db.ai)
+
+- The 60db TTS playback uses [mpv](https://mpv.io/installation/); install it and make sure it is on your PATH
 
 - To run ```main_ws.py``` you will need a [Azure](https://portal.azure.com/) OpenAI Deployment and Speech Service Resource
 
@@ -50,7 +52,7 @@ pip install -r requirements.txt
 cp .env.template .env
 ```
 
-- Edit the `.env` file and add your OpenAI API key, ElevenLabs API key, and desired ElevenLabs voice ID.
+- Edit the `.env` file and add your OpenAI API key plus the keys of the TTS provider you want to use (ElevenLabs or 60db).
 
 ## Usage
 
@@ -74,9 +76,27 @@ cp .env.template .env
 
 ## Customization
 
-You can customize the voice used by the AI Assistant by changing the `ELEVENLABS_VOICE_ID` in the `.env` file.
+### Choosing the TTS provider
+
+Set `TTS_PROVIDER` in the `.env` file to switch between providers:
+
+```bash
+TTS_PROVIDER=elevenlabs   # ElevenLabs (default)
+TTS_PROVIDER=sixtydb      # 60db
+```
+
+Only the API keys of the selected provider are required:
+
+- `elevenlabs`: `ELEVENLABS_API_KEY`, `ELEVENLABS_VOICE_ID`
+- `sixtydb`: `SIXTYDB_API_KEY`, `SIXTYDB_VOICE_ID`
+
+`main.py` uses the 60db TTS stream API (`POST /tts-stream`), `main_ws.py` uses the 60db TTS WebSocket (`wss://api.60db.ai/ws/tts`).
+
+### Changing the voice
+
+You can customize the voice used by the AI Assistant by changing the `ELEVENLABS_VOICE_ID` or `SIXTYDB_VOICE_ID` in the `.env` file.
 
-A link to the pre-made voices is in the `.env.template` file.
+Links to the pre-made voices are in the `.env.template` file.
 
 ## License
 
@@ -97,3 +117,4 @@ If you encounter any issues or have questions, please open an issue on GitHub.
 
 - [OpenAI](https://www.openai.com/)
 - [ElevenLabs](https://www.elevenlabs.io/)
+- [60db](https://60db.ai/)

config.py

@@ -32,10 +32,23 @@
 if OPENAI_SYSTEM_PROMPT is None:
     raise ValueError("OPENAI_SYSTEM_PROMPT not set")
 
-ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
-if ELEVENLABS_API_KEY is None:
-    raise ValueError("ELEVENLABS_API_KEY not set")
+TTS_PROVIDER = os.getenv("TTS_PROVIDER", "elevenlabs").strip().lower()
+if TTS_PROVIDER not in ("elevenlabs", "sixtydb"):
+    raise ValueError("TTS_PROVIDER must be 'elevenlabs' or 'sixtydb'")
 
+ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
 ELEVENLABS_VOICE_ID = os.getenv("ELEVENLABS_VOICE_ID")
-if ELEVENLABS_VOICE_ID is None:
-    raise ValueError("ELEVENLABS_VOICE_ID not set")
+
+SIXTYDB_API_KEY = os.getenv("SIXTYDB_API_KEY")
+SIXTYDB_VOICE_ID = os.getenv("SIXTYDB_VOICE_ID")
+
+if TTS_PROVIDER == "elevenlabs":
+    if ELEVENLABS_API_KEY is None:
+        raise ValueError("ELEVENLABS_API_KEY not set")
+    if ELEVENLABS_VOICE_ID is None:
+        raise ValueError("ELEVENLABS_VOICE_ID not set")
+else:
+    if SIXTYDB_API_KEY is None:
+        raise ValueError("SIXTYDB_API_KEY not set")
+    if SIXTYDB_VOICE_ID is None:
+        raise ValueError("SIXTYDB_VOICE_ID not set")

config_ws.py

@@ -40,10 +40,23 @@
 if AZUREAI_REGION is None:
     raise ValueError("AZUREAI_REGION not set")
 
-ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
-if ELEVENLABS_API_KEY is None:
-    raise ValueError("ELEVENLABS_API_KEY not set")
+TTS_PROVIDER = os.getenv("TTS_PROVIDER", "elevenlabs").strip().lower()
+if TTS_PROVIDER not in ("elevenlabs", "sixtydb"):
+    raise ValueError("TTS_PROVIDER must be 'elevenlabs' or 'sixtydb'")
 
+ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
 ELEVENLABS_VOICE_ID = os.getenv("ELEVENLABS_VOICE_ID")
-if ELEVENLABS_VOICE_ID is None:
-    raise ValueError("ELEVENLABS_VOICE_ID not set")
+
+SIXTYDB_API_KEY = os.getenv("SIXTYDB_API_KEY")
+SIXTYDB_VOICE_ID = os.getenv("SIXTYDB_VOICE_ID")
+
+if TTS_PROVIDER == "elevenlabs":
+    if ELEVENLABS_API_KEY is None:
+        raise ValueError("ELEVENLABS_API_KEY not set")
+    if ELEVENLABS_VOICE_ID is None:
+        raise ValueError("ELEVENLABS_VOICE_ID not set")
+else:
+    if SIXTYDB_API_KEY is None:
+        raise ValueError("SIXTYDB_API_KEY not set")
+    if SIXTYDB_VOICE_ID is None:
+        raise ValueError("SIXTYDB_VOICE_ID not set")

main.py

@@ -1,7 +1,14 @@
 """
-This script uses the OpenAI API and ElevenLab's REST API
+This script uses the OpenAI API and the ElevenLabs REST API
+or the 60db TTS stream API, selected via TTS_PROVIDER.
 """
+import base64
+import json
+import shutil
+import subprocess
 import sys
+
+import requests
 import speech_recognition as sr  # type: ignore
 from openai import OpenAI
 from elevenlabs import stream
@@ -14,8 +21,11 @@
     OPENAI_PROJECT_ID,
     OPENAI_MODEL,
     OPENAI_SYSTEM_PROMPT,
+    TTS_PROVIDER,
     ELEVENLABS_API_KEY,
     ELEVENLABS_VOICE_ID,
+    SIXTYDB_API_KEY,
+    SIXTYDB_VOICE_ID,
 )
 
 console = Console()
@@ -25,14 +35,15 @@
     organization=OPENAI_ORG_ID,
     project=OPENAI_PROJECT_ID,
 )
-e_client = ElevenLabs(api_key=ELEVENLABS_API_KEY)
+if TTS_PROVIDER == "elevenlabs":
+    e_client = ElevenLabs(api_key=ELEVENLABS_API_KEY)
 voice_ident = ELEVENLABS_VOICE_ID
 
 
 def generate_and_play_response(user_input, conversation_history):
     """
-    Generates response using the OpenAI API and
-    plays it using the ElevenLabs TTS API.
+    Generates response using the OpenAI API and plays it using
+    the selected TTS provider (ElevenLabs or 60db).
 
     Args:
         user_input (str): The user's input.
@@ -72,6 +83,10 @@ def generate_and_play_response(user_input, conversation_history):
     def text_stream():
         yield response_text
 
+    if TTS_PROVIDER == "sixtydb":
+        tts_stream_sixtydb(response_text)
+        return
+
     audio_stream = e_client.generate(
         text=text_stream(),
         voice=Voice(
@@ -91,6 +106,78 @@ def text_stream():
     stream(audio_stream)
 
 
+def tts_stream_sixtydb(text):
+    """
+    Converts text to speech using the 60db TTS stream API
+    (POST https://api.60db.ai/tts-stream) and plays the streamed
+    MP3 chunks through the mpv player.
+
+    The response is newline-delimited JSON (NDJSON): each "chunk"
+    message contains a base64-encoded piece of audio, followed by a
+    final "complete" message.
+
+    Args:
+        text (str): The text to convert to speech.
+
+    Returns:
+        None
+
+    Raises:
+        ValueError: If mpv is not installed on the system.
+        RuntimeError: If the 60db API returns an error message.
+    """
+    if shutil.which("mpv") is None:
+        raise ValueError(
+            "mpv not found, necessary to stream audio. "
+            "Install instructions: https://mpv.io/installation/"
+        )
+
+    response = requests.post(
+        "https://api.60db.ai/tts-stream",
+        headers={
+            "Authorization": f"Bearer {SIXTYDB_API_KEY}",
+            "Content-Type": "application/json",
+        },
+        json={
+            "text": text,
+            "voice_id": SIXTYDB_VOICE_ID,
+            "speed": 1,
+            "stability": 50,
+            "similarity": 75,
+        },
+        stream=True,
+        timeout=60,
+    )
+    response.raise_for_status()
+
+    mpv_process = subprocess.Popen(
+        ["mpv", "--no-cache", "--no-terminal", "--", "fd://0"],
+        stdin=subprocess.PIPE,
+        stdout=subprocess.DEVNULL,
+        stderr=subprocess.DEVNULL,
+    )
+
+    try:
+        for line in response.iter_lines():
+            if not line:
+                continue
+            data = json.loads(line)
+            if data.get("type") == "chunk":
+                audio_chunk = base64.b64decode(
+                    data["result"]["audioContent"]
+                )
+                mpv_process.stdin.write(audio_chunk)
+                mpv_process.stdin.flush()
+            elif data.get("type") == "error":
+                raise RuntimeError(
+                    f"60db TTS error: {data.get('message')}"
+                )
+    finally:
+        if mpv_process.stdin:
+            mpv_process.stdin.close()
+        mpv_process.wait()
+
+
 def recognize_speech(timeout=20):
     """
     Recognizes speech from the microphone input.