reverie-sdk 0.1.4

Reverie Language Technologies SDK

Reverie Python SDK

Official Reverie API documentation
PyPI project

Supported Python versons:

• 3.7.x
• 3.8.x
• 3.9.x
• 3.10.x
• 3.11.x
• 3.12.x

Installing

1. Minimal dependencies (excluded service: tts_streaming)

   # total size: approx 150-200 MB
   pip install reverie_sdk[py3x]
   # example: 
   # pip install reverie_sdk[py37]
   

2. Full dependencies

   # total size: approx 1.5-5.5 GB
   pip install reverie_sdk[py3x-all]
   # example: 
   # pip install reverie_sdk[py37-all]
   

where: x = 7, 8, 9, 10, 11 and 12

How to

ASR/STT | stt_stream

Sample 0: From Bytes (Simulated file streaming)

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

for resp in client.asr.stt_stream(
    src_lang="en",
    bytes_or_stream=open("/path/to/audio.wav", "rb").read(),
    continuous="0",
    domain="generic",
    format="16k_int16",
    logging="true",
    punctuate="true",
    silence=15,
    timeout=15,
):
    print(resp)

Sample Output:

ReverieAsrResult (
    id           : xxxxxxa8257a4e45eb
    text         : 
    final        : False
    cause        : partial
    success      : True
    confidence   : 1.0
    display_text : 
)
ReverieAsrResult (
    id           : xxxxxxa8257a4e45eb
    text         : but what if somebody
    final        : False
    cause        : partial
    success      : True
    confidence   : 0
    display_text : but what if somebody
)
                        .
                        .
                        .
ReverieAsrResult (
    id           : xxxxxxa8257a4e45eb
    text         : but what if somebody decides to break and be careful that you keep adequate coverage but look for places to save money maybe it's taking longer to get things squared away then the bankers expected hiring the wife for one's company may win her tenth
    final        : False
    cause        : partial
    success      : True
    confidence   : 0
    display_text : but what if somebody decides to break and be careful that you keep adequate coverage but look for places to save money maybe it's taking longer to get things squared away then the bankers expected hiring the wife for one's company may win her tenth
)
ReverieAsrResult (
    id           : xxxxxxa8257a4e45eb
    text         : but what if somebody decides to break it be careful that you keep adequate coverage but look for places to save money maybe it's taking longer to get things squared away then the bankers expected hiring the ymca for one's company may win her tech cited retirement and
    final        : True
    cause        : silence detected
    success      : True
    confidence   : 0.905
    display_text : But what if somebody decides to break it? Be careful that you keep adequate coverage, but look for places to save money. Maybe it's taking longer to get things squared away. then. the bankers expected hiring the Y.M.CA for one's company may win her tech cited retirement, and.
)

Sample 1: From Stream (Mic)

Sample Code:

import pyaudio, asyncio
from reverie_sdk import ReverieClient
from reverie_sdk.services.asr import AudioStream

client = ReverieClient(
    api_key="REVERIE_API_KEY",
    app_id="REVERIE_APP_ID",
)

stream = AudioStream()
pa = pyaudio.PyAudio()

def mic_callback(in_data, frame_count, time_info, status):
    try:
        asyncio.run(stream.add_chunk_async(in_data))
    except:
        return (None, pyaudio.paAbort)
    return (None, pyaudio.paContinue)

async def main():
    pa_stream = pa.open(
        rate=RATE,
        channels=CHANNELS,
        format=FORMAT,
        frames_per_buffer=CHUNK_SIZE,
        input=True,
        stream_callback=mic_callback,
    )
    pa_stream.start_stream()

    print("listening... press ctrl+C to stop", flush=True)

    try:
        await client.asr.stt_stream_async(
            src_lang="en",
            bytes_or_stream=stream,
            callback=print,
            continuous="0",
            domain="generic",
            format="16k_int16",
            logging="true",
            punctuate="true",
            silence=15,
            timeout=15,
        )
    except Exception as err:
        print(err)
    finally:
        pa_stream.stop_stream()
        pa_stream.close()


asyncio.run(main())

pa.terminate()

Sample Output:

ALSA lib pcm_dsnoop.c:567:(snd_pcm_dsnoop_open) unable to open slave
ALSA lib pcm_dmix.c:1000:(snd_pcm_dmix_open) unable to open slave
ALSA lib pcm.c:2721:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.rear
ALSA lib pcm.c:2721:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.center_lfe
ALSA lib pcm.c:2721:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.side
... (SOME WARNINGS) ...
listening... press ctrl+C to stop
ReverieAsrResult (
    id           : xxxxxxa384ee3c64b3d
    text         : 
    final        : False
    cause        : ready
    success      : True
    confidence   : 1.0
    display_text : 
)
ReverieAsrResult (
    id           : xxxxxxa384ee3c64b3d
    text         : 
    final        : False
    cause        : partial
    success      : True
    confidence   : 1.0
    display_text : 
)
ReverieAsrResult (
    id           : xxxxxxa384ee3c64b3d
    text         : hello
    final        : False
    cause        : partial
    success      : True
    confidence   : 0.9876
    display_text : hello
)
ReverieAsrResult (
    id           : xxxxxxa384ee3c64b3d
    text         : hello hello testing one two three
    final        : False
    cause        : partial
    success      : True
    confidence   : 0.9161
    display_text : hello hello testing one two three
)
ReverieAsrResult (
    id           : xxxxxxa384ee3c64b3d
    text         : hello hello testing one two three
    final        : True
    cause        : silence detected
    success      : True
    confidence   : 0.9065
    display_text : Hello hello Testing 123.
)

ASR/STT | stt_file

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.asr.stt_file(
    src_lang="en",
    data=open("./path/to/audio.wav", "rb").read(),
    domain="generic",
    format="16k_int16",
    logging="true",
    punctuate="true",
)

print(resp)

Sample Output:

ReverieAsrResult (
    id           : xxxxxxxxxaa0b80d34a7b
    text         : but what if ... some make beautiful chairs, cabinets, chests, dollhouses, et cetera.
    final        : True
    cause        : EOF received
    success      : True
    confidence   : 0.9475
    display_text : But what if ... some make beautiful chairs, cabinets, chests, dollhouses, et cetera.
)

ASR/STT | stt_batch

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

for resp in client.asr.stt_batch(
    src_lang="en",
    data=open("/path/to/audio.wav", "rb").read(),
    domain="generic",
    format="16k_int16",
    subtitles=True,
):
    print(resp, flush=True)

print(resp)

Sample Output:

ReverieAsrBatchResult (
    code    : 000
    job_id  : xxxxxxb97e45211716
    message : Success. Request accepted
)
ReverieAsrBatchResult (
    code    : 004
    job_id  : xxxxxxb97e45211716
    status  : processing
    message : Your request is being processed.
)
ReverieAsrBatchResult (
    code    : 004
    job_id  : xxxxxxb97e45211716
    status  : processing
    message : Your request is being processed.
)
ReverieAsrBatchResult (
    code    : 004
    job_id  : xxxxxxb97e45211716
    status  : processing
    message : Your request is being processed.
)
ReverieAsrBatchResult (
    code    : 000
    job_id  : xxxxxxb97e45211716
    status  : completed
    message : Transcript ready.
)
ReverieAsrBatchResult (
    code    : 000
    job_id  : xxxxxxb97e45211716
    message : Transcript ready.
    result  :
    _Transcript(
        transcript          : But what if somebody decides to break it? be ... some make beautiful chairs, cabinets, chests, dollhouses, et cetera.
        original_transcript : BUT WHAT IF SOMEBODY DECIDES TO BREAK IT BE ... SOME MAKE BEAUTIFUL CHAIRS CABINETS CHESTS DOLLHOUSES ET CETERA.
        channel_number      : 1
        subtitles           : "1\n00:00:00,000 --> 00:00:51,000\nBUT WHAT IF SOMEBODY DECIDES TO BREAK IT BE ... SOME MAKE BEAUTIFUL CHAIRS CABINETS CHESTS DOLLHOUSES ET CETERA\n\n"
        words               : 155 words
    )
)

TTS | tts

Sample 0: text

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.tts.tts(
    speaker="en_male",
    text="hello there, how are you?",
    sample_rate=16000,
    format="OPUS",
    pitch=0,
    speed=1,
)

print(resp)

Sample Output:

ReverieTtsResponse(
    format          : OPUS
    sample_rate     : 16000
    audio_bytes     : 8392
    duration        : 1.695 seconds
    kwargs          : {'src': {'text': 'hello there, how are you?', 'ssml': None}}
)

Sample 1: ssml

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.tts.tts(
    speaker="en_male",
    ssml="""
<speak>
<voice name="en_female"> This is an example. </voice>
<voice name="hi_female"> यह एक उदाहरण है </voice>
</speak>
""".strip(),
    sample_rate=16000,
    format="OPUS",
    pitch=0,
    speed=1,
)

print(resp)

Sample Output:

ReverieTtsResponse(
    format          : OPUS
    sample_rate     : 16000
    audio_bytes     : 8392
    duration        : 1.695 seconds
    kwargs          : {'src': {'text': 'hello there, how are you?', 'ssml': None}}
)

TTS | Streaming

Note: py3x-all variant is required to install for this. See Installing section.

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

with open("./big_text.txt", encoding="utf-8") as f:
    text = f.read()

for resp_idx, resp in enumerate(
    client.tts.tts_streaming(
        text=text,
        speaker="en_male",
        max_words_per_chunk=5,
        fast_sentence_fragment=False,
    )
):
    print(f"{resp_idx:08d} {resp.duration:10.3f}")
    resp.save_audio(
        f".path/to/output/{resp_idx:08d}.wav",
        create_parents=True,
        overwrite_existing=True,
    )

Sample Output:

00000000     13.351
00000001      5.422
00000002      9.799
...
00000048      6.014
00000049      9.741
00000050     11.622

T13N | transliteration

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.t13n.transliteration(
    data=["Reverie Language Technologies ltd website address is http://www.reverieinc.com"],
    src_lang="en",
    tgt_lang="hi",
    cnt_lang="en",
    convertOrdinal=True,
    ignoreTaggedEntities=False,
    convertNumber="words",
    abbreviationWithoutDot=False,
    isBulk=False,
    noOfSuggestions=3,
    domain="1",
)

print(resp)

Sample Output:

ReverieT13nResponse(
    responseList    : 
        ReverieT13nResponse.Result(
            apiStatus   : 2
            inString    : Reverie Language Technologies ltd website address is http://www.reverieinc.com
            outString   : 
                रेवरी लैंग्वेज टेक्नोलॉजीस लिमिटेड वेबसाइट एड्रेस इज़ http://www.reverieinc.com
        )
)

NMT | localization

Sample 0: Single tgt_lang

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.nmt.localization(
    data=[
        "Reverie Language Technologies was established in 2009.",
        "The company head office is located in Bangalore.",
    ],
    domain=1,
    src_lang="en",
    tgt_lang=["hi"],
    nmtMaskTerms=["Reverie Language Technologies"],
    nmtMask=True,
    debugMode=True,
    enableNmt=False,
    enableTransliteration=True,
    enableLookup=True,
)

print(resp)

Sample Output:

ReverieLocalizationResult(
    responseList:
        ReverieLocalizationResponseItem(
            inString    = Reverie Language Technologies was established in 2009.
            outString   = रेवेरी लैंगएज टेच्नोलोजीस वैस इस्टैब्लिश्ड इन 2009।
            apiStatus   = 10
        )
        ReverieLocalizationResponseItem(
            inString    = The company head office is located in Bangalore.
            outString   = द कोम्पैनी हीड ओफिस इस लोकैट्ड इन बैंगैलोर।
            apiStatus   = 10
        )
    tokenConsumed: 0
)

Sample 1: Multiple tgt_lang

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.nmt.localization(
    data=[
        "Reverie Language Technologies was established in 2009.",
        "The company head office is located in Bangalore.",
    ],
    domain=1,
    src_lang="en",
    tgt_lang=["hi", "ta"],
    nmtMaskTerms=["Reverie Language Technologies"],
    nmtMask=True,
    debugMode=True,
    enableNmt=False,
    enableTransliteration=True,
    enableLookup=True,
)

print(resp)

Sample Output:

ReverieLocalizationResult(
    responseList:
        ReverieLocalizationResponseItem(
            inString    = Reverie Language Technologies was established in 2009.
            apiStatus   = 4
            outStrings:
                hi : रेवेरी लैंगएज टेच्नोलोजीस वैस इस्टैब्लिश्ड इन 2009.
                ta : ரெவெரி லஙுய்ஜ் டெச்னோலோஜீஸ் வஸ் எஸ்டப்லிஷெட் இன் 2009.
        
        )
        ReverieLocalizationResponseItem(
            inString    = The company head office is located in Bangalore.
            apiStatus   = 4
            outStrings:
                hi : द कोम्पैनी हीड ओफिस इस लोकैट्ड इन बैंगैलोर.
                ta : த் கோம்பனி ஹீட் ஓஃபிஸ் இஸ் லோகடெட் இன் பஙலோர்.
        
        )
    tokenConsumed: 0
)

NLU | lang_id_text

Sample Code:

from reverie_sdk import ReverieClient

client = ReverieClient(
    api_key="MY_API_KEY",
    app_id="MY_APP_ID",
)

resp = client.nlu.lang_id_text(
    "भारत दक्षिण एशिया में स्थित भारतीय उपमहाद्वीप का सबसे बड़ा देश है",
    max_length=16,
)

print(resp)

Sample Output:

ReverieLangIdTextResponse(
    lang        = hi
    confidence  = 0.9999784231185913
)

Integration Status

1. Speech [✔️]
   1. stt_stream [✔️]
   2. stt_file (non-streaming) [✔️]
   3. stt_batch (async) [✔️]
2. NMT [✔️]
   1. localization [✔️]
3. T13N[✔️]
   1. transliteration [✔️]
4. TTS [✔️]
   1. tts [✔️]
   2. tts streaming [✔️]
5. NLU [ ]
   1. lang_id_text [✔️]
   2. train_model [ ]
   3. test_model [ ]

Note:

1. All functions have basic validations in place, atleast datatype checks