Hi,
Anybody have any success with Voice Recognition on the phone without an internet, as in downloading the analysis repository or something?
I've Been looking for the same for a long time now
If you are into developing apps this could help.
Building application with pocketsphinx (sorry can't post links yet, but just google it)
PocketSphinx is a small-footprint continuous speech recognition system, freely licensed under a simplified BSD license, suitable for handheld and desktop applications. It features:
Cross-platform: Linux, Windows, Mac OS X, iPhoneOS
Experimental support for Nokia S60v3 and Windows Mobile
Support for semi-continuous, phonetically-tied, and fully continuous acoustic models
Model footprint on disk of about 10MB per language
Memory footprint under 20MB for medium-vocabulary continuous recognition
Trigram language models and JSGF finite-state grammars
Acoustic models for English and Mandarin
Small language models for English and Mandarin (simplified and traditional characters)
Python language bindings
GStreamer multimedia framework integration
Have fun
Dip
Xxxdevilxxx said:
Hi,
Anybody have any success with Voice Recognition on the phone without an internet, as in downloading the analysis repository or something?
Click to expand...
Click to collapse
Looking for this, now that this is available in 4.1. Tried this in nexus s, not only is this offline - it's way more accurate also! Offline accuracy is absolute credo compared to this.
I really hope this can be ported to galaxy s2.
hirak99 said:
Looking for this, now that this is available in 4.1. Tried this in nexus s, not only is this offline - it's way more accurate also! Offline accuracy is absolute credo compared to this.
I really hope this can be ported to galaxy s2.
Click to expand...
Click to collapse
Hi
I want to know about how to start offline voice recognition system in android 5.1 system update it was working in my old android kikat 4.4.4 and even I can also call by just saying call and the person name but now none is working. Do you have any idea how to use offline speech recognition system. I have already download the English UK offline pack and also tts pack for it and also tell me how to use hands free in it
Related
I think the Samsung TTS is much better than Pico one. Unfortunately, in my S2 i only have English(USA), Chinese and Korean. Did anyone know from (if) where we can download more voices - particular interested in English(UK)?
Indeed. It's much better. Looking for Spanish myself.
//EDIT: In a related note, could it be possible to use Samsung's TTS in Navigation? Right now setting it as default produces gibberish in Navigation (see this thread). I'd really like to have, say, Spanish Samsung TTS in Navigation, or even English if nothing else. Pico sucks. :-(
I think Google Navigation wont "speak" if the language is different from phone language. So if the phone is English(UK) and Google Navigation is English(USA) isn't good.
Is it possible to download/purchase additional voices for Samsung TTS? If no, then Samsung TTS is useless for me.
There are however SVOX TTS in Android Market: Classic Text To Speech Engine.
This TTS engine have voices in 25 languages available for purchase and I have read that SVOX have excellent voice quality.
hello...
Just bump into this thread while I was looking for a free & good TTS enginer.
DJViking, you're right - SVOX and many others (such as Ivona) have better "voice" - it is much less robotized that Tico TTS. I'm right now downloading Ivona.. it's free to 15th Sept (note: the language UK file is as big as 250 mb)
Good luck with the search -- now I'm trying to figure out where's the Samsung TTS in my phone. It disappear when I install Ficeto JVQ ROM on my galaxy S
Bump up!
There's no additional language pack for stock Samsung TTS?
Hi guys,
we will have to buy a new phone for my visually impaired sister during next year (or maybe much earlier), because she is currently unable to use her simple Nokia (unable to read SMS messages on display etc.).
I tried to find any simple phone with voice output for her, but the only one possible candidate was Samsung Haven, which was only available for U.S. mobile networks and not for european GSM (and did not have Czech language support, but this was minor in this case).
She needs only a simple phone to perform standard voice calls and SMS messages. Current approach here is to use some smartphone – symbian with Mobilespeak or Android with Code Factory or some other a11y suite. OK, why not, but this aims to make a accessible smartphone, while many people need only accessible phone (which is not available).
So this také me to following idea:
if Android uses a modified linux kernel and GUI via Java virtual machine Dalvik, then it should be possible to use alternative simple menu with something like linux console voice output instead of the standard GUI and so make it much quicker.
This menu could be very easy – only to provide functions similar to „normal“ cellphones like the old good Nokia 3310 only with voice output. I know that it sounds very stupid to degrade a smartphone to a simple cellphone, but I do not know other way.
I think, that such trimmed off custom ROM with this sort of text menu and only basic functions (calls, SMS, phone book, calendar and clock with alarm) combined with a voice output could be able to run on cheap and relatively weak devices like Huawei Boulder for example so final solution could be comparable with normal cellphones. We need something with physical keyboard, so we could use HTC Cha Cha or if possible, the much cheaper Huawei which are both available from official distribution.
Are there any initiatives or attempts to create something similar?
Thanks much
PS: we are ready to sacrifice wireless internet and lack of Google services for better performance.
Please use the Q&A Forum for questions Thanks
Moving to Q&A
so in android 4.0.1 a new ''updated" version of the android TTS pico engine was introduced. it sounds much less robotic now. In Scott's clean ROM the new voice has been implemented, but in Newt's 4.0.3 ROM the old pico voice is used.
Does anyone know which files control the TTS engine? I already tried replacing the /system/tts folder in newts rom with the one from clean rom and it didnt work.
I know there are a few apps in the market to change the tts voice, but i dont want to pay the $3. i would rather donate it to the rom developers here on xda
this should really be posted in general, as it is not development and a question, though i understand you wanted to get more developer attention by posting here.
That being said, here is some news i found here:
http://developer.android.com/sdk/android-4.0.html
Text-to-speech Engines
Android’s text-to-speech (TTS) APIs have been significantly extended to allow applications to more easily implement custom TTS engines, while applications that want to use a TTS engine have a couple new APIs for selecting an engine.
Using text-to-speech engines
In previous versions of Android, you could use the TextToSpeech class to perform text-to-speech (TTS) operations using the TTS engine provided by the system or set a custom engine using setEngineByPackageName(). In Android 4.0, the setEngineByPackageName() method has been deprecated and you can now specify the engine to use with a new TextToSpeech constructor that accepts the package name of a TTS engine.
You can also query the available TTS engines with getEngines(). This method returns a list of TextToSpeech.EngineInfo objects, which include meta data such as the engine’s icon, label, and package name.
Building text-to-speech engines
Previously, custom engines required that the engine be built using an undocumented native header file. In Android 4.0, there is a complete set of framework APIs for building TTS engines.
The basic setup requires an implementation of TextToSpeechService that responds to the INTENT_ACTION_TTS_SERVICE intent. The primary work for a TTS engine happens during the onSynthesizeText() callback in a service that extends TextToSpeechService. The system delivers this method two objects:
SynthesisRequest: This contains various data including the text to synthesize, the locale, the speech rate, and voice pitch.
SynthesisCallback: This is the interface by which your TTS engine delivers the resulting speech data as streaming audio. First the engine must call start() to indicate that the engine is ready to deliver the audio, then call audioAvailable(), passing it the audio data in a byte buffer. Once your engine has passed all audio through the buffer, call done().
Now that the framework supports a true API for creating TTS engines, support for the native code implementation has been removed. Look for a blog post about a compatibility layer that you can use to convert your old TTS engines to the new framework.
For an example TTS engine using the new APIs, see the Text To Speech Engine sample app.
Click to expand...
Click to collapse
The bolded part leads me to believe that you need new framework api(s) to be able to change over. Though i could be completely wrong, this was just my best guess.
Also, since we do not have a FULL release, but rather a ICS leak, it may be that there are more difficulties than a full release would have.
i was going to post in general, but like you said i thought the devs would have a better chance at seeing it here. a mod can move if necessary.
i did see the post you mentioned on a google developer site. i guess the new APIs will be implemented when we finally have a official ICS build.
you should have went with your gut and post it in the general section. the devs will see it, or hear about it or someone who has knowledge will answer it. please keep the dev section clean for development. thanks.
Used the same google email addr I had in my old evo 4g. The issue is that the voice translation doesn't seem to be understanding most words I say. Had no such issues with the old evo, in fact, close to 90%+ accuracy. Cleared the voice history from my google profile (link from the google dashboard), but the issue continues.
I don't use speech-text often, but it would be nice if it worked. Everything else works on the phone, so its likely some software issue.
Thoughts?
Cleared the voice history from my google profile (link from the google dashboard),
Click to expand...
Click to collapse
this.
now google-ri has to relearn your speaking. give it a couple weeks and it should get better again.
I assumed that was the right thing to do, given that this is a new piece of hardware, probably 'hears' me differently than the old one.
How does the google voice program learn from it's mistakes. It's not as though it lets you correct the incorrectly identified words.
its hidden in the "privacy policy" (yea i know, i'm weird for reading all that stuff)
What is the benefit of personalized speech recognition?
Speech recognition is based on statistical modeling. To recognize spoken words, we compare the input speech to a statistical model of the language and try to find the closest match - the system's best guess at what the user said. The statistical model is huge - it must cover all of the fundamental sounds of the language (phonemes), all of the words, and all of the different ways that the words can be strung together in the spoken language. Furthermore, it must capture all of the variations among users that happen when a language is spoken, for example all of the different dialects and accents and individual differences in the sound of the voice (e.g., male vs. female, young vs. old).
Knowing what you said in the past allows us to build specialized models that are designed to match your voice and your words. Over time, this allows us to improve the speech recognition accuracy for you. But to do that, we must keep track of what you said in the past.
Click to expand...
Click to collapse
A note about corrections
When you speak, we present to you what it is that our algorithm thought you said. When you make a correction we capture that correction. However, we do it only for words - no numbers or words with special characters. Furthermore, we limit the correction only to cases where you substitute one word with one or two other words. If you simply add text we do not capture it.
Click to expand...
Click to collapse
---------- Post added at 06:01 AM ---------- Previous post was at 06:00 AM ----------
its tied to your google account, not the hardware.
Hey all,
So I'm running the new Google Experience Launcher (GEL) on CyanogenMod 10.1, but I'm a little annoyed by the lack of hotword support in UK English. If I set my language to English (US), hotword support DOES WORK. However, the voice recognition is less accurate (because I'm British) and the synthesised voice is American instead of British. Etc. Etc.
Now I see no good reason why Google can't enable hotword search for all languages. But they haven't. And rather than sit around feeling sad, I thought I'd fix it myself!
My plan is for an XPosed module to hook any neccesary methods, and supply modified code that enabled hotword searches.
I've begun reverse engineering the Google Search APK, where all the actual GEL code lives (from the 3.1.8 apk, available here)
I started out looking at the smali, but with something as complex as this, it soon got tiresome.
I'm now looking at the decompiled classes.dex I got out of dex2jar. However, despite finding multiple methods that reference hotword support, I can't seem to really trace it all the way back. The only condition I've found so far is a check as to whether the device is a low-ram device.
Just wondering if any veterans can point me in the right direction? This is a big hacking project for me, and I reckon, a great learning experience.
I'd host the decompiled code for convenience, but I expect that'd be considered rather naughty.
I also don't understand, why Google didn't support the hotword search for languages that work with voice search.
I'm very interested in this, if someone finds what to patch. I'm using german language
Also, why ONLY "Ok Google"?
We should have choices what the hot word should be.
Maybe I want to say "Hey, Slave" or "Computer!" (Star trek...lol...)
Nevertheless, good luck!