smallestai 1.2.0__py3-none-any.whl → 1.3.1__py3-none-any.whl

smallest/async_tts.py

@@ -6,21 +6,21 @@ from typing import Optional, Union, List
 
 from .models import TTSModels, TTSVoices
 from .exceptions import TTSError, APIError
-from .utils import (TTSOptions, validate_input, preprocess_text, add_wav_header,
-                     get_smallest_languages, get_smallest_voices, get_smallest_models, API_BASE_URL)
+from .utils import (TTSOptions, validate_input, preprocess_text, add_wav_header, split_into_chunks,
+                     get_smallest_languages, get_smallest_voices, get_smallest_models, SENTENCE_END_REGEX, API_BASE_URL)
 
 
 class AsyncSmallest:
     def __init__(
-            self,
-            api_key: Optional[str] = None,
-            model: TTSModels = "lightning",
-            sample_rate: int = 24000,
-            voice: TTSVoices = "emily",
-            speed: Optional[float] = 1.0,
-            add_wav_header: Optional[bool] = True,
-            transliterate: Optional[bool] = False,
-            remove_extra_silence: Optional[bool] = False
+        self,
+        api_key: Optional[str] = None,
+        model: TTSModels = "lightning",
+        sample_rate: int = 24000,
+        voice: TTSVoices = "emily",
+        speed: Optional[float] = 1.0,
+        add_wav_header: Optional[bool] = True,
+        transliterate: Optional[bool] = False,
+        remove_extra_silence: Optional[bool] = False
     ) -> None:
         """
         AsyncSmallest Instance for asynchronous text-to-speech synthesis.
@@ -47,7 +47,8 @@ class AsyncSmallest:
         """
         self.api_key = api_key or os.environ.get("SMALLEST_API_KEY")
         if not self.api_key:
-            raise TTSError("API key is required")
+            raise TTSError()
+        self.chunk_size = 250
         
         self.opts = TTSOptions(
             model=model,
@@ -70,6 +71,7 @@ class AsyncSmallest:
         if self.session:
             await self.session.close()
 
+
     def get_languages(self) -> List[str]:
         """Returns a list of available languages."""
         return get_smallest_languages()
@@ -110,42 +112,45 @@ class AsyncSmallest:
             setattr(opts, key, value)
 
         validate_input(text, opts.voice, opts.model, opts.sample_rate, opts.speed)
-
-        payload = {
-            "text": preprocess_text(text),
-            "sample_rate": opts.sample_rate,
-            "voice_id": opts.voice,
-            "add_wav_header": opts.add_wav_header,
-            "speed": opts.speed,
-            "model": opts.model,
-            "transliterate": opts.transliterate,
-            "remove_extra_silence": opts.remove_extra_silence
-        }
-
-        headers = {
-            "Authorization": f"Bearer {self.api_key}",
-            "Content-Type": "application/json",
-        }
-
-        if not self.session:
-            self.session = aiohttp.ClientSession()
+        chunks = split_into_chunks(text)
+        audio_content = b""
+
+        for chunk in chunks:
+            payload = {
+                "text": preprocess_text(chunk),
+                "sample_rate": opts.sample_rate,
+                "voice_id": opts.voice,
+                "add_wav_header": False,
+                "speed": opts.speed,
+                "model": opts.model,
+                "transliterate": opts.transliterate,
+                "remove_extra_silence": opts.remove_extra_silence
+            }
+
+            headers = {
+                "Authorization": f"Bearer {self.api_key}",
+                "Content-Type": "application/json",
+            }
+
+            if not self.session:
+                self.session = aiohttp.ClientSession()
         
-        async with self.session.post(f"{API_BASE_URL}/{opts.model}/get_speech", json=payload, headers=headers) as res:
-            if res.status != 200:
-                raise APIError(f"Failed to synthesize speech: {await res.text()}. For more information, visit https://waves.smallest.ai/")
+            async with self.session.post(f"{API_BASE_URL}/{opts.model}/get_speech", json=payload, headers=headers) as res:
+                if res.status != 200:
+                    raise APIError(f"Failed to synthesize speech: {await res.text()}. For more information, visit https://waves.smallest.ai/")
             
-            audio_content = await res.read()
+                audio_content += await res.read()
 
         if save_as:
             if not save_as.endswith(".wav"):
                 raise TTSError("Invalid file name. Extension must be .wav")
             
-            if self.opts.add_wav_header:
-                async with aiofiles.open(save_as, mode='wb') as f:
-                    await f.write(audio_content)
-            else:
-                async with aiofiles.open(save_as, mode='wb') as f:
-                    await f.write(add_wav_header(audio_content, self.opts.sample_rate))
+            async with aiofiles.open(save_as, mode='wb') as f:
+                await f.write(add_wav_header(audio_content, self.opts.sample_rate))
+
             return None
 
+        if opts.add_wav_header:
+            return add_wav_header(audio_content, self.opts.sample_rate)
+        
         return audio_content

smallest/exceptions.py

@@ -1,6 +1,9 @@
 class TTSError(Exception):
     """Base exception for TTS SDK"""
-    pass
+    default_message = "API key is required. Please set the `SMALLEST_API_KEY` environment variable or visit https://waves.smallest.ai/ to obtain your API key."
+    
+    def __init__(self, message=None):
+        super().__init__(message or self.default_message)
 
 class APIError(TTSError):
     """Raised when the API returns an error"""

smallest/stream_tts.py

@@ -34,31 +34,52 @@ class TextToAudioStream:
             max_retries: Number of retry attempts for failed synthesis (default: 3)
         """
         self.tts_instance = tts_instance
+        self.tts_instance.opts.add_wav_header = False
+
         self.sentence_end_regex = SENTENCE_END_REGEX
         self.queue_timeout = queue_timeout
         self.max_retries = max_retries
         self.queue = Queue()
         self.buffer_size = 250
         self.stop_flag = False
-        self.tts_instance.opts.add_wav_header = False
 
 
     async def _stream_llm_output(self, llm_output: AsyncGenerator[str, None]) -> None:
         """
-        Streams the LLM output, splitting it into sentences and adding each to the queue.
+        Streams the LLM output, splitting it into sentences based on the regex 
+        and chunk size, and adding each chunk to the queue.
 
         Parameters:
         - llm_output (AsyncGenerator[str, None]): An async generator yielding LLM output.
         """
         buffer = ""
+        last_break_index = 0
+
         async for chunk in llm_output:
             buffer += chunk
-            if self.sentence_end_regex.match(buffer) or self.buffer_size > 600:
-                self.queue.put(buffer)
-                buffer = ""
+            i = 0
+
+            while i < len(buffer):
+                current_chunk = buffer[:i + 1]
+                if self.sentence_end_regex.match(current_chunk):
+                    last_break_index = i
+
+                if len(current_chunk) >= self.buffer_size:
+                    if last_break_index > 0:
+                        self.queue.put(buffer[:last_break_index + 1].replace("—", " ").strip())
+                        buffer = buffer[last_break_index + 1:] 
+                    else:
+                        # No sentence boundary, split at max chunk size
+                        self.queue.put(buffer[:self.buffer_size].replace("—", " ").strip())
+                        buffer = buffer[self.buffer_size:] 
+
+                    last_break_index = 0
+                    i = -1 
+
+                i += 1
 
         if buffer:
-            self.queue.put(buffer)
+            self.queue.put(buffer.replace("—", " ").strip())
 
         self.stop_flag = True  # completion flag when LLM output ends
 

smallest/tts.py

@@ -6,20 +6,20 @@ from typing import Optional, Union, List
 
 from .models import TTSModels, TTSVoices
 from .exceptions import TTSError, APIError
-from .utils import (TTSOptions, validate_input, preprocess_text, 
-get_smallest_languages, get_smallest_voices, get_smallest_models, API_BASE_URL)
+from .utils import (TTSOptions, validate_input, preprocess_text, add_wav_header, split_into_chunks,
+get_smallest_languages, get_smallest_voices, get_smallest_models, SENTENCE_END_REGEX, API_BASE_URL)
 
 class Smallest:
     def __init__(
-            self,
-            api_key: Optional[str] = None,
-            model: TTSModels = "lightning",
-            sample_rate: int = 24000,
-            voice: TTSVoices = "emily",
-            speed: Optional[float] = 1.0,
-            add_wav_header: Optional[bool] = True,
-            transliterate: Optional[bool] = False,
-            remove_extra_silence: Optional[bool] = True
+        self,
+        api_key: Optional[str] = None,
+        model: TTSModels = "lightning",
+        sample_rate: int = 24000,
+        voice: TTSVoices = "emily",
+        speed: Optional[float] = 1.0,
+        add_wav_header: Optional[bool] = True,
+        transliterate: Optional[bool] = False,
+        remove_extra_silence: Optional[bool] = True
     ) -> None:
         """
         Smallest Instance for text-to-speech synthesis.
@@ -45,7 +45,9 @@ class Smallest:
         """
         self.api_key = api_key or os.environ.get("SMALLEST_API_KEY")
         if not self.api_key:
-            raise TTSError("API key is required")
+            raise TTSError()
+        
+        self.chunk_size = 250
         
         self.opts = TTSOptions(
             model=model,
@@ -57,6 +59,7 @@ class Smallest:
             transliterate=transliterate,
             remove_extra_silence=remove_extra_silence
         )
+    
         
     def get_languages(self) -> List[str]:
         """Returns a list of available languages."""
@@ -99,41 +102,49 @@ class Smallest:
 
         validate_input(text, opts.voice, opts.model, opts.sample_rate, opts.speed)
 
-        payload = {
-            "text": preprocess_text(text),
-            "sample_rate": opts.sample_rate,
-            "voice_id": opts.voice,
-            "add_wav_header": opts.add_wav_header,
-            "speed": opts.speed,
-            "model": opts.model,
-            "transliterate": opts.transliterate,
-            "remove_extra_silence": opts.remove_extra_silence,
-        }
-
-        headers = {
-            "Authorization": f"Bearer {self.api_key}",
-            "Content-Type": "application/json",
-        }
+        chunks = split_into_chunks(text)
+        audio_content = b""
+
+        for chunk in chunks:
+            payload = {
+                "text": preprocess_text(chunk),
+                "sample_rate": opts.sample_rate,
+                "voice_id": opts.voice,
+                "add_wav_header": False,
+                "speed": opts.speed,
+                "model": opts.model,
+                "transliterate": opts.transliterate,
+                "remove_extra_silence": opts.remove_extra_silence,
+            }
+
+            headers = {
+                "Authorization": f"Bearer {self.api_key}",
+                "Content-Type": "application/json",
+            }
+
+            res = requests.post(f"{API_BASE_URL}/{opts.model}/get_speech", json=payload, headers=headers)
+            if res.status_code != 200:
+                raise APIError(f"Failed to synthesize speech: {res.text}. Please check if you have set the correct API key. For more information, visit https://waves.smallest.ai/")
+            
+            audio_content += res.content
+
 
         res = requests.post(f"{API_BASE_URL}/{opts.model}/get_speech", json=payload, headers=headers)
         if res.status_code != 200:
             raise APIError(f"Failed to synthesize speech: {res.text}. Please check if you have set the correct API key. For more information, visit https://waves.smallest.ai/")
-        
-        audio_content = res.content
 
         if save_as:
             if not save_as.endswith(".wav"):
                 raise TTSError("Invalid file name. Extension must be .wav")
             
-            if self.opts.add_wav_header:
-                with open(save_as, "wb") as wf:
-                    wf.write(audio_content)
-            else:
-                with wave.open(save_as, "wb") as wf:
-                    wf.setnchannels(1)
-                    wf.setsampwidth(2)
-                    wf.setframerate(self.opts.sample_rate)
-                    wf.writeframes(audio_content)
+            with wave.open(save_as, "wb") as wf:
+                wf.setnchannels(1)
+                wf.setsampwidth(2)
+                wf.setframerate(self.opts.sample_rate)
+                wf.writeframes(audio_content)
             return None
-            
+        
+        if self.opts.add_wav_header:
+            return add_wav_header(audio_content, self.opts.sample_rate)
+    
         return audio_content

smallest/utils.py

@@ -11,7 +11,7 @@ from .models import TTSModels, TTSLanguages, TTSVoices
 
 
 API_BASE_URL = "https://waves-api.smallest.ai/api/v1"
-SENTENCE_END_REGEX = re.compile(r'.*[-.!?;:…\n]$')
+SENTENCE_END_REGEX = re.compile(r'.*[-.—!?;:…\n]$')
 SAMPLE_WIDTH = 2
 CHANNELS = 1
 
@@ -53,11 +53,56 @@ def preprocess_text(text: str) -> str:
     # Replace special characters with their normal form
     text = unicodedata.normalize('NFKD', text).encode('ASCII', 'ignore').decode('ASCII')
     text = text.lower()
+    text = text.replace("—", " ")
     # Normalize punctuation using Moses punct normalizer
     mpn = MosesPunctNormalizer()
     text = mpn.normalize(text)
     return text.strip()
 
+def split_into_chunks(self, text: str) -> List[str]:
+        """
+        Splits the input text into chunks based on sentence boundaries 
+        defined by SENTENCE_END_REGEX and the maximum chunk size.
+        """
+        chunks = []
+        current_chunk = ""
+        last_break_index = 0
+
+        i = 0
+        while i < len(text):
+            current_chunk += text[i]
+
+            # Check for sentence boundary using regex
+            if SENTENCE_END_REGEX.match(current_chunk):
+                last_break_index = i
+
+            if len(current_chunk) >= self.chunk_size:
+                if last_break_index > 0:
+                    # Split at the last valid sentence boundary
+                    chunk = text[:last_break_index + 1].strip()
+                    chunk = chunk.replace("—", " ")
+                    chunks.append(chunk)
+
+                    text = text[last_break_index + 1:]
+                    i = -1  # Reset index to process the remaining text
+                    current_chunk = ""
+                    last_break_index = 0
+                else:
+                    # No sentence boundary found, split at max length
+                    current_chunk = current_chunk.replace("—", " ")
+                    chunks.append(current_chunk.strip())
+                    text = text[self.chunk_size:]
+                    i = -1  # Reset index to process the remaining text
+                    current_chunk = ""
+
+            i += 1
+
+        if text:
+            text = text.replace("—", " ")
+            chunks.append(text.strip())
+
+        return chunks
+
 
 def get_smallest_languages() -> List[str]:
     return list(TTSLanguages.__args__)

{smallestai-1.2.0.dist-info → smallestai-1.3.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: smallestai
-Version: 1.2.0
+Version: 1.3.1
 Summary: Official Python client for the Smallest AI API
 Author-email: Smallest <info@smallest.ai>
 License: MIT
@@ -18,12 +18,12 @@ Requires-Dist: requests
 Requires-Dist: sacremoses
 Requires-Dist: pydub
 Provides-Extra: test
-Requires-Dist: jiwer ; extra == 'test'
-Requires-Dist: httpx ; extra == 'test'
-Requires-Dist: pytest ; extra == 'test'
-Requires-Dist: pytest-asyncio ; extra == 'test'
-Requires-Dist: deepgram-sdk ; extra == 'test'
-Requires-Dist: python-dotenv ; extra == 'test'
+Requires-Dist: jiwer; extra == "test"
+Requires-Dist: httpx; extra == "test"
+Requires-Dist: pytest; extra == "test"
+Requires-Dist: pytest-asyncio; extra == "test"
+Requires-Dist: deepgram-sdk; extra == "test"
+Requires-Dist: python-dotenv; extra == "test"
 
 ![image](https://i.imgur.com/TJ2tT4g.png)   
 
@@ -102,7 +102,20 @@ if __name__ == "__main__":
 - `speed`: Speech speed multiplier (default: 1.0)
 - `add_wav_header`: Include WAV header in output (default: True)
 - `transliterate`: Enable text transliteration (default: False)
-- `remove_extra_silence`: Remove additional silence (default: True)
+- `remove_extra_silence`: Remove additional silence (default: True)  
+
+These parameters are part of the `Smallest` instance. They can be set when creating the instance (as shown above). However, the `synthesize` function also accepts `kwargs`, allowing you to override these parameters for a specific synthesis request.
+
+For example, you can modify the speech speed and sample rate just for a particular synthesis call:  
+```py
+client.synthesize(
+    "Hello, this is a test for sync synthesis function.",
+    save_as="sync_synthesize.wav",
+    speed=1.5,  # Overrides default speed
+    sample_rate=16000  # Overrides default sample rate
+)
+```
+
 
 ### Async   
 Asynchronous text-to-speech synthesis client.    
@@ -134,7 +147,18 @@ if __name__ == "__main__":
 - `speed`: Speech speed multiplier (default: 1.0)
 - `add_wav_header`: Include WAV header in output (default: True)
 - `transliterate`: Enable text transliteration (default: False)
-- `remove_extra_silence`: Remove additional silence (default: True)
+- `remove_extra_silence`: Remove additional silence (default: True)  
+
+These parameters are part of the `AsyncSmallest` instance. They can be set when creating the instance (as shown above). However, the `synthesize` function also accepts `kwargs`, allowing you to override any of these parameters on a per-request basis.
+
+For example, you can modify the speech speed and sample rate just for a particular synthesis request:  
+```py
+audio_bytes = await tts.synthesize(
+    "Hello, this is a test of the async synthesis function.",
+    speed=1.5,  # Overrides default speed
+    sample_rate=16000  # Overrides default sample rate
+)
+```
 
 ### LLM to Speech    
 

smallestai-1.3.1.dist-info/RECORD

@@ -0,0 +1,12 @@
+smallest/__init__.py,sha256=vaoIBml_IobavpVvFazB86iikg2iEy4h3ddxqv_0Fy4,190
+smallest/async_tts.py,sha256=ReYTePzR0e3UcWxwnetGnwO4q_F7g2LRZPxIVfmgB4Y,6120
+smallest/exceptions.py,sha256=nY6I8fCXe2By54CytQ0-i3hFiYtt8TYAKj0g6OYsCjc,585
+smallest/models.py,sha256=R5UZZA9SibrJ2DsWPi_mkKI13WfyC-MLd-7kptfjns4,390
+smallest/stream_tts.py,sha256=0OypcUzgP7CN3VGcGJDnQ2FDw2JOzPaSQ1cXK69k5dY,6198
+smallest/tts.py,sha256=l8VHaOE8-Feg3Ey8C3osOrLs3ffYz0q_J1ACiEtZ8y0,5999
+smallest/utils.py,sha256=Xg4sYqhTgEgmT6H0qMNmlVlF1ilqY2BUCBTlk-_yONg,3795
+smallestai-1.3.1.dist-info/LICENSE,sha256=kK3HNKhN7luQhkjkNWIvy9_gizbEDUM4mSv_HWq9uuM,1068
+smallestai-1.3.1.dist-info/METADATA,sha256=anPPrjMmnWa_b1S65Wg23efwLUaoww7pbyU2qLhZTAk,9853
+smallestai-1.3.1.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+smallestai-1.3.1.dist-info/top_level.txt,sha256=i5ktbWkG-2aS28vrYTeuhKtA-tY2ZG7SHgLHi87QTLw,9
+smallestai-1.3.1.dist-info/RECORD,,

{smallestai-1.2.0.dist-info → smallestai-1.3.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.5.0)
+Generator: setuptools (75.6.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

smallestai-1.2.0.dist-info/RECORD

@@ -1,12 +0,0 @@
-smallest/__init__.py,sha256=vaoIBml_IobavpVvFazB86iikg2iEy4h3ddxqv_0Fy4,190
-smallest/async_tts.py,sha256=w_SY1Oetn5Zorq-8JXA7lGeRHR3kTtBzqotc_hF0hOQ,6010
-smallest/exceptions.py,sha256=41GLVvNTfRQMQsPLGk0lHuhK2mak8_dVtiFLEtT23Dc,333
-smallest/models.py,sha256=R5UZZA9SibrJ2DsWPi_mkKI13WfyC-MLd-7kptfjns4,390
-smallest/stream_tts.py,sha256=1j4JpAwrAmwprC98mKQwuhXf0HFxFTlMcZ3_JAdcAK0,5416
-smallest/tts.py,sha256=2v6kBPC_hocWdvt2Uua2aQgzg3BKQGIFXpq2DZMz9aA,5540
-smallest/utils.py,sha256=hAgyEfZEnvayzu8qS4LXhpZR8qK7z4gatLWGVOkS3Yg,2183
-smallestai-1.2.0.dist-info/LICENSE,sha256=kK3HNKhN7luQhkjkNWIvy9_gizbEDUM4mSv_HWq9uuM,1068
-smallestai-1.2.0.dist-info/METADATA,sha256=qsNxKNN-KNTshjeBBh9AhS_yeYzlHEqPlt_9zuV2fGY,8717
-smallestai-1.2.0.dist-info/WHEEL,sha256=R06PA3UVYHThwHvxuRWMqaGcr-PuniXahwjmQRFMEkY,91
-smallestai-1.2.0.dist-info/top_level.txt,sha256=i5ktbWkG-2aS28vrYTeuhKtA-tY2ZG7SHgLHi87QTLw,9
-smallestai-1.2.0.dist-info/RECORD,,