Try Fish Audio for free 👉 https://fish.audio/?fpr=pritam71
I Built My Own AI Voice Assistant with Fish Audio S2.1 Pro API
Build your own AI voice assistant with Fish Audio S2.1 Pro API! In this hands-on tutorial, I demonstrate text-to-speech (TTS), API integration, natural AI voice generation and emotion tags that change how your assistant speaks. Explore how expressive speech can bring LLM-powered AI agents, virtual assistants, AI tutors and avatars to life—with no local GPU setup. Watch the voice demos, follow the API walkthrough and start building your own voice-enabled application. What personality would you give your AI assistant? Comment below and subscribe to PritamSahooAI for practical AI tutorials.
#fishaudio #S21Pro #AIVoiceAssistant #TextToSpeech #AITutorial #FishAudioTurns1
I Built My Own AI Voice Assistant with Fish Audio S2.1 Pro API
Build your own AI voice assistant with Fish Audio S2.1 Pro API! In this hands-on tutorial, I demonstrate text-to-speech (TTS), API integration, natural AI voice generation and emotion tags that change how your assistant speaks. Explore how expressive speech can bring LLM-powered AI agents, virtual assistants, AI tutors and avatars to life—with no local GPU setup. Watch the voice demos, follow the API walkthrough and start building your own voice-enabled application. What personality would you give your AI assistant? Comment below and subscribe to PritamSahooAI for practical AI tutorials.
#fishaudio #S21Pro #AIVoiceAssistant #TextToSpeech #AITutorial #FishAudioTurns1
Category
🛠️
LifestyleTranscript
00:00what if your ai assistant didn't just know what to say but actually knew how to say it excited calm
00:07serious tired even laughed today i am going to build my own voice assistant using an api
00:13and here is the interesting part the voice layer i am using is currently free no gpu complex setup
00:20no hosting a voice model let's build it so now before we get start coding if you haven't heard
00:26about fish audio here is the simplest way i can explain it think of fish audio as an ai voice
00:32engine for creators and developers you can give it a text or a response generated by an ai and fish
00:40audio turns into that expressive spoken audio so if you use it for things like text to speech voice
00:48cloning ai assistant avatars applications or even your own agent workflows the important difference
00:54for today's video is that i am not interested in simply typing something into a website
00:58or downloading mp3 i want voice assistant in my own application you can easily build it
01:04and integrate it my ai generates the intelligence and fish audio gives the intelligence a voice
01:11and that is sounding similar to me and it is that exactly we are going to build using s2.1
01:18pro api
01:19that the latest release fish audio describes itself as a professional ai audio platform built for
01:26creators and developers with s2.1 pro api focused on making ai speech more expressive that's the bottom
01:35line hey if you have not met me my name is pritam and on this channel i don't want to
01:39just tell you
01:39another ai model is launched i want to see what actually we can build it quick disclosure this video
01:46is sponsored by fish audio but instead of doing another boring work through home page settings
01:52pricing done right i want to solve a real problem i want my own ai assistant that you can give
01:58an answer
01:58and speak the answer naturally so here is the architecture so simple and before that before that
02:05meet s2.1 pro for this experiment i am using fish audio s2.1 pro currently it is free developer
02:13model
02:13string is s2.1 pro free according to the documentation it supports 83 languages using one
02:21model quite insane and the current free api has no hard usage cap under its fair usage policy
02:28but the feature that really caught my attention is emotion tax instead of just adjusting 20 technical
02:34parameters you can directly say to the ai you would direct a voice actor for example laughing excited
02:43or even sign right you can use more descriptive instructions so let's stop talking and let's
02:49share it out and let's build the demo to keep it simple i have kept it intentionally very simple
02:55you can use your own voice and let's get started before setting an api hop over to fish.audio i
03:03will
03:03give link in the description to check for that and before that you need to record a voice of yours
03:08right
03:09and to record a voice simply click my voice and build new voice and you can set instant voice clone
03:20or if
03:20you want a professional setup it will require some time one to two hours better for professional gets
03:26studio quality sound but for the demo i will use this one demo i will use instant voice clone
03:31and click continue either you can upload or record so let me record it live right so let me start
03:38it
03:39okay this ai has gone insane and recent news of anthropic researcher resigning and saying ai will wipe
03:46out the entire human base in next 10 years sounds too far-fetched right if asi comes that is artificial
03:55super intelligence but i would say that's not the reality the reality is we have just started entering
04:02the agi era with projects like astra from open ai so let's analyze this audio and continue to details
04:12and you can name pritamsahu ai new s2 dot one let's say and click create and you need to accept
04:25it you can
04:25make it private unlisted i will go public that's completely okay if somebody wants to use it now
04:31your step one is done now let's see how can you use your own voice and you can play it
04:37around right
04:38let me go to creative or developer and let's say you want to go creative you can do that way
04:44but i will
04:45use developer one as promised using an api just one thing i am already having fish audio api key i
04:52will
04:52show you how to do it but be careful never expose the api key in your code i am storing
04:58in my environment
05:00right and i will be using this to call fish audio text to speech api using s2.1 pro so
05:09first thing
05:09first let's start this one on the api key come over to developers here right and check for api key
05:19and you can create one i have been creating and playing around this you can see it here
05:24and you can name it let's say demo sample let's say i say seven days and create it right and
05:33i have
05:33saved my key let's say for example and then what you can do is you can go to api playground
05:40so this is
05:41one of the easy way if you are coming from a non-coder background right but i promised i will
05:47say developer
05:47background right and here is the key where all the things emotions get added you can use here s21 pro
05:55directly but i will be using command line and show you the developer way right so make sure you have
06:01proper model here right reference id here which is quite critical and the output what you want to name
06:09it right and just run let's say welcome to fish audio this is my first generated voice let's run the
06:15request and you can also use python and javascript using curl here in shortly in the command line but
06:23let's say it in case you are not comfortable with the coding hello welcome to fish audio
06:30this is my first ai generated voice you can see the emotions being added right this quite critical and
06:38i named my voice as pritam saw ai news 2.1 what you can do is you can shift now
06:45to the terminal right
06:48so i have set all the parameters let me zoom out a bit right and something like i want to
06:56say be careful
06:58exposing the apis especially right the first command you need to run is this one you can copy it which
07:04you saw
07:05in your screen write the call command and simply place once you have set the environment variable
07:10and reference let me press this and run it in front of you live right so now it is done
07:18and my output
07:20file is welcome dot mp3 first ai generated voice hello welcome to fish audio this is my first ai generated
07:31voice i just added the emotions right and it is matching my voice so you saw that right how easy
07:38it is you can use the web console as well as command line tool right as long as you give
07:43reference id
07:44and set the keys properly for apis and this is you are no longer saying or changing what ai says
07:52we are
07:53changing how ai performs and we pushed one further step right this is because i found it interesting
08:00instead of thinking like a sound engineer i started writing my instructions so let me do this and change
08:06it more let's say let's say i are surprised right let me join the text to show you how easy
08:14it is
08:16to get it working so i said the ai companies are racing to build leaders at open ai and x
08:22i have
08:23backed the pacing of ai advancement over safety concerns right make sure you have reference id
08:29and the format and let me say i use welcome one this time to make sure like i have a
08:35proper understanding
08:37of the output file let me go back the screen and clear it out and paste this command right no
08:45sweat it
08:46is available out of the box in your console we saw it right and the output is already generated quite
08:51fast
08:52inside we play this out the companies racing to build powerful ai are now calling for the race to slow
08:58down leaders at anthropic open ai and x ai have back facing advanced ai development over safety concerns
09:07it's insanely exactly with my voice right we saw that right how realistic you can make that actual ai
09:13assistant but here is the important part i don't want to manually type everything so imagine give me my
09:19morning brief in 15 seconds ai generates you have priorities you finish the customer proposal prepare
09:25for the partner meeting and review your project roadmap now my application could add that energetic
09:31professional before sending the answer to fish audio but imagine it's 11 pm i ask summarize my day now i
09:39can
09:39send relaxed calmly or imagine there is an urgent issue the application will use the
09:46all the seriousness the empathy says right all required now think what just happened the context can
09:53determine the voice and all your emotional angles that much more powerful that ai assistant can generate
09:59text to speech where every response sounds exactly the same in your known natural tone and this
10:06architecture doesn't have to stop here demo you can easily integrate this with your own application you
10:12could build an ai tutor that changes the tone when explaining difficult concepts and customer
10:17assistant that sounds calm when somebody is frustrated an ai avatar a game character onboarding assistant
10:24or a voice interface inside your sas product right this audio brief specifically calls out agent
10:31pipelines building around llm plus speech to text plus text to speech along with the use cases including
10:37voice agents avatars games and edtech these where the apis become more interesting than a simple voice
10:45generator because voice becomes your new infrastructure and what i like and my fair understanding these are the
10:52few things i like about this approach first it is programmable i am not logged into somebody else user
10:58interface second the emotional controls the free natural i am describing how i want things to be spoken
11:05instead of adjusting endless technical settings and third for developer experimenting with voice
11:11applications the currently s2.1 pro free api makes prototyping very easier the documentation clearly says
11:19the free access a limited time cal is no sls or guaranteed latency but it is worth the test because
11:25it works amazingly great and the primary position for experimentation and prototyping just a disclaimer it
11:33also advises the developers to check privacy commercial use terms before the production deployment right
11:38you should not be using somebody else voice the and let me call out the free promotion runs till end
11:44of november 2026 so definitely check the latest terms if you are watching the video later and here is my
11:51big
11:51message and take away from this experiment for last few years we have been focused on heavily on teaching ai
11:57what to say i think the next legging layer is teaching ai how to say because the assistance will not
12:03sound
12:04exactly the same when it is congratulating you warning you teaching you or giving something you morning
12:10briefing right voice isn't just the output anymore voice is becoming part of the ai interface and apis like
12:18this makes the interface programmable very easy if you have a non-developer you get the command and you just
12:24copy
12:24paste it in the terminal that's it so here is my challenge for you don't just generate another ai
12:29voice over build something build your own assistant give your coding a voice agent build an ai teacher
12:36create an avatar and take one idea that you have been sitting and connect to fish audio s2 2.1
12:42pro api
12:43and i have added the link in the description and the pin comment and i want to answer one question
12:48below
12:49if your ai assistant have your personality what would you build a teacher a coach a personal
12:56assistant or something completely lazy or crazy if you enjoy watching this video and we actually build
13:02instead of just and hype subscribe to my channel pritam sao ai and this video was sponsored by fish audio
13:09i am thankful to them and see you in the next one